Appendix G
1
Quality Management Analysis by Failure Mode Rank
2
3
4
5
This appendix lists each FM developed during the FMEA, and describes relevant QM that can
6
help mitigate those failures. This appendix provides an example of the potential for customizing
7
and optimizing a QM program based on a specific clinic’s workflows and equipment. The long-
8
term goal is to help create a risk-informed IMRT QM program for each department and process.
9
10
Starting with the highest-ranking RPN failure mode, example strategies for mitigation of each
11
failure are discussed below. Note that many of the QM steps developed in this listing are
12
summarized (and organized) in the example checklists in the main body of the report.
13
Rank
#1
RPN
388
Step#
31
Process
Step
4. Other Pre-Treatment Imaging 6. Images correctly
for CTV Localization
interpreted
FM: Incorrect interpretation of tumor or normal tissue.
14
15
See text in Sec 9 C2 in the main body of TG 100 protocol.
16
Rank
#2
RPN
366
Step #
58
Process
Step
7. RTP
Delineate GTV/CTV (MD) and other
Anatomy
structures
FM: > 3σ contouring error, wrong organ, site, or expansions
17
18
See text in Sec. 9 C3 in the main body of TG 100 protocol.
19
Rank
#3
RPN
354
Step#
209
Process
Step
12. Day N
Tx delivered
Tx
FM: Linac hardware failures: /wrong dose/MU; MLC leaf motions inaccurate,
flatness/symmetry, energy, etc.
20
21
Below we expand on the discussion of this important failure mode presented in Sec. 9 C4 in the
22
main body of TG 100 protocol. The third-highest risk failure mode includes all delivery errors
1
23
associated with failures of the treatment unit – that is, deviations of the linac radiological or
24
geometric performance parameters from their expected values. The FMEA assessed the risk of
25
this FM under the assumption that no periodic device-QA is performed. Because many such FMs
26
are essentially undetectable without such QA until they cause clinical harm, a very high RPN
27
number results, underscoring the importance of periodic QA checks to reduce the risk of linac
28
hardware failures.
29
30
Currently accepted measures for mitigating linac hardware failures are based upon the
31
recommendations of AAPM TG 401 and, more recently, those of TG1422, which supplements
32
and extends the TG 40 approach to modern linacs with newer modalities and accessories, such as
33
multileaf collimators (MLC) and onboard imaging systems. The stated aims of TG 40 and 142
34
are that individual parameter tolerances and QA test frequencies should be such that, when all
35
contributing geometric and dosimetric uncertainties are summed in quadrature, the total
36
cumulative dose-delivery uncertainty will not exceed 5% or 5 mm. These tolerances originated in
37
AAPM Report 13,3 which also provides simple examples making it clear that setup and tissue
38
motion errors, as well as dose-computation errors, are included in the 5 mm or 5% uncertainty
39
budget. In practice both TG 40 and TG142 recommendations are based on TG member
40
consensus and not formal error propagation analyses. Below, TG-100 suggests a simple model
41
that describes a way of assigning tolerances and frequencies based on assessment of risk. It is
42
important to note that additional data are required to make this simple method robustly
43
applicable to the clinic.
44
45
Risk-based considerations suggest that meeting the TG-40 and TG-142 tolerances and
46
measurement frequency schedule may be neither necessary nor sufficient to protect the patient
47
from “wrong dose” or “wrong position” errors.
48
treatment (IGRT) is used for all patients.
49
significantly smaller than that assumed by Report 13, and the role of ODI, light field, and cross
50
hair accuracy is less important. In that case, the TG-142 recommendations for these parameters
51
may be too stringent, inefficiently using QA resources. On the other hand, following TG-142
52
guidelines may not always be sufficient to meet quality standards.
53
recommends that MLC leaf positioning error be assessed monthly. If meeting this performance
Suppose that daily online image-guided
Actual setup and tissue-motion errors will be
2
For example, TG-142
54
goal really is required to avoid exceeding the upper bound on allowable total dose-delivery
55
uncertainty, theoretically a significant performance deviation could subject a cohort of patients to
56
erroneous doses for an entire treatment course, especially for courses which are less than 30
57
days.
58
indicates the importance of assuring that the maximum number of fractions a linac performance
59
deficit could remain undetected never results in a patient treatment course exceeding the allowed
60
cumulative positional and dose delivery tolerances.
The very high RPN for hardware related failures in the “treatment delivered” step
61
62
TG-100 approach to determination of test frequencies and tolerances
63
Using dose output (Gy/MU in reference geometry, or Dref ) as an example, we illustrate the
64
approach to setting tolerances and test frequencies.
65
66
67
1. Define the QM goal in terms of the overall dose delivery or positional accuracy goals
consistent with the department’s vision of acceptable quality
68
69
This goal could take many forms, all of which involve conservatively estimating worst-case
70
outcome over a typical patient’s course of treatment. An example description of a catastrophic
71
event avoidance criterion would be “no patient shall ever receive a treatment that alters NTCP or
72
TCP from the planned values by more than 20%.”
73
catastrophe), we will use the following goal: “no patient’s total dose-delivery uncertainty should
74
exceed 5%”. The value of 5% has been used for a long time , most often to describe the
75
uncertainty of the patient’s dose relative to the prescription. AAPM Report 134 specifies that the
76
overall uncertainty is to be 2 standard deviations while TG-40 and TG-142 are not specific on
77
this point.
For quality erosion (as opposed to
78
79
80
2. Determine the sensitivity of the relevant QM goals to the linac performance parameter
in question
81
82
Other sources of total dosimetric uncertainty must be accommodated within the allowed 5% total
83
dose uncertainty, referred to here as Dtot , so the effect on the dose uncertainty from all the
84
other parameters, pi, must first be calculated or estimated to determine the remaining partial
3
85
uncertainty that can be allotted to uncertainties in machine output, D(D ref ) . The approximate
86
uncertainty available for treatment-unit output is given by
87
(
)
2
DDtot d Dref = DDtot
-
é
( )
éDD p é .
é tot i é
2
(E 1)
All other
parameters,
iéoutpur
88
For purposes of illustration, assume that the overall dose accuracy goal is ∆Dtot=5% with
89
a 98% confidence (3SD) and the dose computation engine, including other central axis
90
dosimetric parameters (depth-dose ratios, output factors, and attenuator transmission, and others)
91
have a combined uncertainty of 3.0% with a 95% confidence (2 SD). Equation E1 gives
92
DDtot (d Dref ) ~0.8% as 1 SD, or approximately 2.4% at the 3 SD level. At a 95% confidence level
93
of 1.6%, this falls between the recommendation for output consistency for monthly and annual
94
checks in the report of TG 1423. Because a change in output scales all doses by the same
95
fraction, the sensitivity of the total delivered dose to output error is 1.00 so that
96
DDtot Dref = DDref . Such a simplification is not applicable to many other machine performance
97
endpoints.
98
 D
uncertainty ∆q, can be estimated by Dtot  q   Max  tot
 q
99
endpoints that require non-trivial sensitivity analyses (that have not, in general, been performed)
100
include output constancy and linearity as functions of dose rate, gantry angle, and MU/segment.
101
These parameters can affect total dose delivery accuracy but have sensitivities less than one
102
because their dosimetric impact depends on the distribution of gantry angles, MU/segment, and
103
dose rates characteristic of typical plans.
104
sensitivity analyses, but TG100 has not performed this ambitious project. It remains an area for
105
future research.4,5
( )
Generally, dosimetric sensitivity to an arbitrary machine parameter, q, with an

 q . Examples of performance

The literature contains a few examples of such
106
107
108
109
3. Determine the maximum error in the linac performance endpoint for which the
machine remains operable
110
Determination of the upper bound on erroneous function would require an analysis of the
111
machine control and interlock systems.
The presence of linac interlocks, which prevent
112
operation of the machine when important parameters go out of tolerance, are an important
4
113
consideration in deciding what testing frequency to require; however, the failure of symmetry
114
interlocks highlights, such machine controls reduce the likelihood, but cannot eliminate the
115
possibility, of failure. For the purposes of discussion, we will consider two situations:
116
1. First, assume that transient and persistent dose output errors as large as 40% are possible
117
without triggering machine interlocks that would prevent machine operation. Although we do
118
not expect a modern accelerator interlock system to allow such a large error, such errors have
119
been reported following service when a pot was misadjusted and the interlocks reset to new
120
values. Therefore, this large, but not impossible, value is chosen for demonstration purposes.
121
2. Second, assume a more typical interlock limit of 5%.
122
123
124
4. Determine the monitoring frequency needed to achieve the overall uncertainty goal
125
Given the total uncertainty in the linac performance endpoint of interest that is consistent with
126
clinical goals, a typical number of fractions per course of treatment, and the maximum deviation
127
in performance consistent with linac operation, Eq 1 in the main text of TG100 protocol (Sec. 9
128
C 4a) can be used to determine the number of fractions between monitoring intervals that can be
129
tolerated without exceeding the upper bound on total dose-delivery uncertainty in the assumed
130
worst case scenario with the output error persisting for the entire interval between successive
131
tests (see “Fig. 7”). We emphasize that this model is highly simplified. For example,
132
radiobiological effects due to fraction size changes are not considered. Also, we assume that all
133
treatments on this machine are delivered in the stated number of daily fractions (35, 10 and 5
134
fractions in Fig. 7) and once the error is corrected, all other fractions are without error. For 35
135
fractions under these conditions (solid red curve, Fig. 7), if interlocks allow a 40% error and one
136
allows a dose output error of 5% for each patient’s total course, sampling dose output on every
137
fourth fraction would be satisfactory compared to the daily output monitoring recommended by
138
TG-142. For typical palliative treatment regimens (e.g. 30 Gy in 10 daily fractions), this
139
sampling frequency implies that the corresponding output error could be at worst 12% per
140
treatment (solid black curve in Fig. 7.) As the number of fractions of the patient’s treatment
141
decreases, the number of fractions in which an error can be tolerated decreases, and more
142
frequent checks of the output become more important. To maintain a 5% total dose output error
143
in a 10 fraction treatment in the face of a 40% error in the output would require output checks
5
144
every 1.25 treatments - in reality this would indicate daily checks. On a machine used to deliver
145
single-fraction treatments, output checks prior to the delivery of each single fraction for each
146
new patient might be indicated in the absence of other considerations that lead to confidence that
147
the linac output does not change during a single treatment day.
148
149
5. Establishing action levels and thresholds
150
151
The above analysis by itself offers little insight into establishing action levels or tolerances. If
152
deviations from performance endpoints such as dose output, are less than the corresponding
153
sensitivity-adjusted uncertainty, DD0 , the overall uncertainty goal will be met. The purpose of
154
the system outlined above is not to control the average or likely performance outcome, but to
155
protect the average patient against “outliers” that embody worst-case scenarios, regardless of
156
how unlikely spontaneous occurrence of a persistent 40% error might be.
157
protecting patients against low-probability worst-case scenarios, QA should also seek to
158
minimize overall mean uncertainty of dose delivery. Thus, equipment and calibration protocols
159
should be selected so as to minimize the total uncertainty of the measurement. Selection of the
160
action level, e.g., the error above which the MU integrator is readjusted, should be based upon
161
this consideration. For linac performance endpoint measurements that exhibit significant random
162
variability but within a range well below the clinically observable severity level (S  5), process
163
control charts6,7 and other statistical techniques8 can be used to distinguish underlying trends
164
from day-to-day statistical fluctuations and set a more appropriate action threshold than a simple
165
fixed thresholds.
In addition to
166
167
Other dosimetric and geometric performance endpoints
168
169
1.
Energy and Beam Flatness/Symmetry: Selection of a sampling frequency requires
170
both a sensitivity analysis and assessment of how much deviation from flatness, symmetry or
171
central axis depth dose an operable linac could exhibit. For typical Varian machines, it would
172
seem unlikely that deviations larger than 10% could spontaneously occur in any of the above
173
three items without concomitant failure of machine output or a sustained effort by engineering
174
staff to retune the machine to operate at the wrong energy. While the following is a simplified
6
175
example, a similar approach in the clinic would result in a reasonable recommendation. Suppose
176
that the limit for central axis and off-axis ratio dose delivery errors is Dtot  DCAX , Dprof   4%
177
, where DCAX is the deviation from the clinically used depth dose and Dprof is the deviation
178
from the proper beam profile; further, assume that the combined allowed error could be 4%.
179
Also assume that the sensitivity of total dose delivery error to the performance metrics
180
DCAX and Dprof is 0.5, so an error in these parameters would produce a dosimetric error that is
181
only half as large (a10% deviation would produce a 5% dose error). To keep the total allowed
182
dosimetric error below 4%, the 5% error could only persist for 4/5 of the number of fractions, or
183
28 of 35 fractions. This suggests that monitoring flatness and energy on an approximately
184
monthly basis, would eliminate the possibility of >4% dose delivery errors due to flatness/depth
185
dose deviations for patients receiving at least 35 fractions.
186
187
TG100 considered how such energy and off-axis profile checks should be performed efficiently,
188
for the purposes both of preventing severe and more frequent, quality-eroding dose delivery
189
errors. The current practice often consists of performing separate periodic checks of each key
190
dosimetric characteristic (flatness, symmetry, depth dose, output factor, wedge factor, etc). From
191
the TG-100 perspective, periodic direct measurements of fundamental dosimetric quantities
192
solely for QA are wasted effort.
193
validated, structured but non-specific tests that can check for changes in underlying parameters,
194
e.g., photon energy spectrum, should be sufficient. For example, verifying the constancy of the
195
largest field shallow depth beam profiles is a highly sensitive check of all beam characteristics
196
that are sensitive to source energy, including depth dose energy-sensitive parameters9. A new
197
AAPM task group (TG-198) was recently formed to review and suggest tests to implement the
198
TG-142 recommendations.
Once initial commissioning has occurred and has been
199
200
2. Geometric parameters: MLC and Jaw Calibration and Operation: Without periodic
201
QA checks of MLC operation, failure to irradiate part of the target or to block part of an organ at
202
risk through an unnoticed failure of an MLC is more likely than a large shift in machine output.
203
The common practice of relying on time-consuming patient-specific fluence maps for MLC
204
verification in addition to periodic QA tests designed to comprehensively span the range of
205
clinical practice at a rationally designed frequency may be neither sufficient nor necessary to
7
206
prevent high-severity random MLC failures or smaller quality-degrading changes. This practice
207
assumes that: i) the MLC shapes used for each patient cover the range of leaves and positions
208
used in all patients; ii) the sensitivity of the plans tested during one day or week is adequate to
209
detect problems with MLC function that could affect other patient plans already under treatment;
210
and iii) the frequency of new patient-specific IMRT QA provides sufficient MLC checks to
211
prevent MLC delivery errors from persisting long enough to exceed the geometric or dosimetric
212
accuracy goals over a course of IMRT. None of these assumptions is necessarily true. MLC leaf
213
positioning error may relate to a single leaf or a shift of entire leaf carriage. MLC errors may be
214
caused by (i) leaf calibration errors (a 1 mm leaf gap error results in 5% dose delivery errors for
215
segments with 2 cm average gap size,10
216
excessive high spatial frequency content in prescribed intensity distributions that drives the MLC
217
to or beyond the limits of its mechanical capabilities; or (iv) failure of system to compensate for
218
gravity or gantry angle effects.
(ii)
component wear, such as motor fatigue; (iii)
219
220
One of the first steps in establishing a rational MLC QM procedure entails determining the
221
sensitivity of overall QM goals to deviations in MLC performance metrics. The small number of
222
relevant published studies11,5 show that 1-3 mm positioning errors distributed randomly amongst
223
single leaf pairs have little effect on overall dose delivery accuracy while systematic leaf gap
224
calibration or carriage positioning errors affecting the entire leaf bank can significantly influence
225
delivery accuracy. For dynamic MLC (dMLC) IMRT techniques, Rangel et al11 demonstrated
226
that 1-mm systematic leaf-carriage errors could causes changes in tumor generalized equivalent
227
uniform dose (gEUD) of 3% (prostate) to 6% (head and neck). For step-and-shoot (sMLC)
228
IMRT, Yan et al5 showed that 2-mm systematic leaf errors produced statistically significant
229
changes in the fraction of test points in single IMRT fields with passing gamma or distance to
230
agreement (DTA) values12,5. For both dMLC and sMLC, 1-mm systematic errors can give rise to
231
5% or 5-mm plan delivery errors.10,12
232
positioning error consistent with linac operation is 5 mm and that the magnitude of dose delivery
233
errors linearly increases with systematic MLC error, Eq 1 of Sec. 9 C 4a in the main body of TG
234
100 protocol predicts that approximately weekly MLC QM tests are needed to ensure no patient
235
treated with 30 fractions of IMRT experiences total dose delivery errors exceeding 5% due to
If we assume that the maximum systematic leaf
8
236
MLC leaf errors. The details of appropriate MLC QA tests vary by manufacturer and system
237
design, so there are many tests suggested in the literature13-16.
238
239
3.
Other operating parameters: Other machine operating parameters, e.g., radiation vs.
240
mechanical isocenter coincidence and excursion could be analyzed in a similar fashion. TG100
241
encourages use of non-specific yet highly sensitive tests of changes in machine geometry
242
integrity, e.g., split-field test16 rather than more time consuming direct measurements of isocenter
243
coincidence.
244
245
Summary of Treatment Machine QM
246
247
Table XII suggests frequencies for various treatment unit QA tests that are compatible with the
248
analysis for IMRT in this report. For parameters that require daily testing, it is straightforward to
249
include those tests in the morning QA performed in most departments. Measurements at a depth
250
in phantom screen for changes in output, beam energy, flatness and symmetry. Marking the
251
phantom with appropriate field sizes, can provide a collimator size verification check.
252
253
Real-Time QA During Treatment Delivery
254
255
In addition to periodic treatment machine QA, it is crucial to actively monitor the actual
256
treatment delivery in multiple ways. Automated checks of dosimetry, MLC motion, patient
257
setup and motion, and other issues during each fraction would help maintain accurate delivery.
258
However, currently, the ability of treatment machines to completely and independently monitor
259
their treatment delivery while presenting that confirmation to the therapists is very incomplete,
260
and is an area that deserves significant developmental effort. Several academic institutions have
261
developed in-house monitoring software, so such capabilities are technologically very feasible
262
and should be made a high priority item for commercial development by linac and treatment
263
management system vendors. Technologically simple methods are also important. Occurrence
264
of many types of important delivery errors can be minimized if the therapists carefully monitor
265
the treatment while in progress and feel empowered to pause treatment and consult with an on-
266
site physicist. Examples of such monitoring include verifying that the MLC moves during IMRT
9
267
treatment, either between SMLC segments or dynamically during DMLC deliveries, and that the
268
patient does not move. Attention to the treatment is a major function of the therapists, but
269
distractions or preoccupation provide distractions and opportunities for attention lapses. The
270
QM system should be designed to expect that such lapses will occur.
271
Rank
#4
RPN
333
Step#
48
Process
Step
6. Initial Tx Planning Retreatment, previous treatment,
Directive
brachytherapy, etc.
FM: Other treatments: Wrong summary or not documented
272
273
The next ranked hazard involves missing information or incorrect conclusions about previous
274
treatments that cause the physician to set the initial goal for treatment planning (i.e. the
275
“treatment planning directive”) incorrectly.
276
treatments (neo-adjuvant or concomitant chemotherapy, or a planned surgical tumor removal)
277
that must be coordinated with the RT course. The “Initial Treatment Planning Directive” FTA
278
(see Appendix E) reveals several basic error pathways that lead to either the wrong summary of
279
other treatments or missing or incorrect documentation of other treatments: (a) the physician
280
fails to take an adequate patient history or review available records; (b) information available for
281
reconstructing past treatments is incorrect; (c) erroneous interpretation or reconstruction of prior
282
treatment from available information; (d) an important prior treatment remains hidden to the
283
current caregivers because no information is available. These error pathways may result from
284
systemic failures to provide necessary information or to supply personnel or time resources
285
adequate to gather complete information, or to human failures such as inattention or drawing
286
incorrect conclusions. The following anecdote from a physician (AJM) reveals how difficult it
287
can be to detect these events (in this case, a potential failure of type d). When directly asked by
288
the physician, a thoracic RT patient denied having prior RT. It turned out that the patient
289
believed his prior RT (for Hodgkin’s disease) to be irrelevant to his current treatment because it
290
was administered many years ago. Only when the patient’s wife appreciatively observed that the
291
current RT experience was much more pleasant than the patient’s prior experience did the
292
physician intercept this serious event. The ideal solution to such problems is a complete and
293
computer-searchable patient medical record in an international medical record database. More
294
practical present-day measures include assertive questioning of the patient during the
By extension, this could also include other
10
295
consultation and ensuring that consultation reports are shared with referring physicians. Because
296
many of the failure modes here are not amenable to automation, level 3 tools from Table III must
297
be used (“protocols, standards and information”). These could include a paper or electronic
298
check-list for the physician to use at consultation, an institutional policy requiring stringent
299
efforts to obtain outside medical records and review of reconstructions of previously
300
administered dose distributions by a physicist.
301
302
To ensure that other treatments are properly coordinated with the RT course, site-specific
303
treatment protocols or checklists should be used for every patient so that there is a standard
304
outline of the appropriate multimodality treatment plan. Example Checklist 1 in Table IV
305
outlines elements of an example site-specific treatment protocol, which can also be configured as
306
a checklist, and lists most of the key elements normally specified during the initial planning
307
directive. Directions that deviate from the standard should be noted by crossing out the standard
308
direction and writing in the desired patient-specific instruction. Such a checklist provides the
309
basis from which all downstream team members involved with the treatment can check the
310
current plan, including physician-driven and technical parameters, against the site-specific
311
treatment process protocol that was selected by the physician. While the dosimetrists and
312
medical physicist can use the protocol as QC for certain physician decisions, there are many
313
steps (e.g., appropriateness of selected protocol, modifications driven by prior RT, etc.) that can
314
only be checked by physician peer review.
315
316
This discussion illustrates two additional key prerequisites for an effective QM program
317

QM for an ongoing planning and delivery process requires a clear expectation of what
318
constitutes a “normal” outcome at each process step; that information must be communicated
319
to the entire staff, e.g. by means of check lists such as the examples given in Tables IV-VIII.
320

Patients with similar clinical presentations should be treated in as similar a fashion as
321
possible, such that exceptions are unusual and not the rule. Necessary exceptions should be
322
well documented.
323
Successful implementation of these strategies requires physicians to surrender, or at least temper,
324
some of their decision-making autonomy. Only if physicians and other staff buy in to such a
325
scheme, and the institution vigorously enforces this expectation, will such strategies be effective.
11
326
Rank
#5
RPN
326
Step#
59
Process
Step
7. RTP
Delineate GTV/CTV and other structures
Anatomy
FM: Excessive delineation errors resulting in <3σ segmentation errors
327
328
This failure mode is similar to Rank #2, but leads to bad results, rather than very bad ones. QM
329
issues are same as described under Rank #2. There is an additional potential cause because the
330
resulting smaller errors may be more difficult to detect, namely “Availability of defective
331
material/tools/equipment.” This cause results from a failure to remove defective equipment from
332
the work environment, which may be a human failure in the maintenance process or to
333
managerial decisions (e.g., not providing sufficient maintenance support for the equipment.)
334
Preventing such failures requires a managerial commitment to provide adequate resources for the
335
facility to achieve its mission. A peer review QA at the end of the RTP Anatomy step would
336
serve as the main QM procedure to cover this potential failure.
337
Rank
#6
RPN
316
Step#
65
Process
Step
7. RTP
PTV Construction
Anatomy
FM: Margin width protocol for PTV construction is inconsistent with actual
distribution of patient setup errors
338
339
One reason for the occurrence of this error is that many centers do not assess consistency
340
between margins and the actual setup errors. Though the reasons for this error in PTV
341
construction are different than the earlier target problems (Ranks 2 and 5), the main QM issues
342
are the same. A check of the PTV and the margin information used for its construction should be
343
included in the anatomy QA check (Example Checklist 2 in Table V) to mitigate this failure
344
mode.
345
Rank
#7
RPN
313
Step#
137
346
Process
Step
9. Plan
1. Plan OK to go to Treatment
Approval
FM: Bad plan approved
In this FM, the physician selects a poor plan, either by erroneously selecting a different and
347
inferior plan than that intended by the dosimetrist, or by approving a plan that is inferior for other
12
348
reasons. With no further review of the quality of the chosen plan, the poor plan will be prepared
349
and exported to the delivery system. The FTA shows that failures with similar effects on
350
patients occur with other failure modes: correct plan approved by physician but wrong plan
351
exported to the delivery system (Step 135, Rank 20) and approving and exporting a plan
352
belonging to another patient to the current patient’s delivery system treatment record (Step 138,
353
Rank 65). One underlying cause, miscommunication, can be mitigated by structuring the plan
354
approval process so that: (a) the same dosimetrist who performs planning also staffs post-
355
planning activities; (b) dosimetrist and physician together select the plan to be approved; or (c)
356
on a particularly tough case, the dosimetrist seeks guidance from another dosimetrist or physicist
357
to improve the ultimate plan quality. Another approach involves creation of a checklist
358
comparing each prescribed goal and constraint to the corresponding result in the approved plan.
359
Setting the criteria for acceptable discrepancies should be a joint effort between the radiation
360
oncologist and the medical physicist or dosimetrist, but once the criteria are established, they
361
provide a framework for detecting a “bad plan.” A significant discrepancy should trigger an alert
362
that the current plan is suboptimal, and an investigation into the reasons. At times, approval of
363
what might generally be considered an inferior plan occurs because the acceptability criteria
364
have been altered as a necessary compromise; if so, this should be documented. There are also
365
times when a better plan is reasonably feasible but a genuinely bad plan results from upstream
366
contouring or constraint specification errors or from inadequate time for planning or from
367
planner inexperience: if such a case is identified, a new plan should be generated. The site-
368
specific protocol and physics plan check (Example Checklists 1in Table IV and 3 in Table VI)
369
can be used to evaluate the physician’s choice of plan by comparing the plan goals with the final
370
results achieved by the plan. Because the severity of undetected bad plan implementation is
371
high, an overall QA check of plan quality is recommended. This check should be part of the
372
physicist’s independent pretreatment plan review (generally regarded as standard of practice, per
373
TG-40) and involves a comparison between the written dose prescription, the selected treatment
374
protocol, and the approved plan results.
375
Rank
#8
RPN
310
Step#
205
Process
Step
12. Day N Treatment machine and peripheral hardware
Tx
setup for Tx
FM: Special motion management methods (e.g. gating, breath-hold) not applied
13
or incorrectly applied
376
377
Failures related to motion management can arise from several causes, each of which requires its
378
own actions. The following address the causes listed in the FMEA:
379

Poorly designed software or hardware. Problems with the equipment and its software
380
should be found during commissioning of the motion management system or process. In
381
general, however, any system should be watched carefully for situations not anticipated
382
and, thus, not tested during commissioning, so that new uses of the system can be
383
commissioned, characterized, and its failure modes studied and addressed.
384

385
386
Use by inexperienced persons.
Sophisticated motion management techniques are
relatively new and appropriate staff training is essential for safe and effective clinical use.

Operator not reacting to information presented by the system.
Failure to react
387
appropriately to information from the system that indicates a problem (e.g. highly
388
irregular breathing, breathing being out of sync with the treatment delivery) poses a
389
major problem, not only for the use of motion management devices, but also for any
390
activities requiring real-time monitoring.
391
appropriate training, the most common cause for inaction when action is required is
392
inattention.
This becomes a problem with any routine task that demands constant
393
monitoring.
It has many causes (e.g., lack of sleep, long work shifts, distractions,
394
boredom) and few effective steps exist to prevent the problem. Assigning two persons to
395
monitor the process provides little added security because the second person tends to rely
396
on the first and loses attention also. The system design must account for human factors
397
and the mix of operator and technical functions. But any system that relies on operator
398
attention must recognize that occasionally attention will fail. This report does not make a
399
definitive recommendation on QM for motion management because the issue is currently
400
the topic of much research.
401

Assuming that the operator received
Failure to engage the system. This is similar to other failures where actions are not
402
forced by interlocks, such as neglecting to insert blocks or place bolus. The frequency of
403
such omissions decreases as treatment unit computers increasingly include checks on
404
more aspects of the treatment. For example, wedge angle and orientation has almost
405
disappeared as a failure mode since interlocks prevent use of the wrong wedge/wedge
14
406
direction as long as the plan was correctly loaded into the delivery system. Similarly,
407
omission of the breathing motion management system will not occur if the treatment unit
408
computer is configured to prevent delivery until the breathing system is engaged. Thus,
409
adequately addressing this cause requires a system tied to the treatment unit computer.
410
411
While not listed with motion management in the FMEA, another possible failure is using an
412
overly optimistic PTV/ITV margin. This is a special case of the Rank #6 failure. Minimizing
413
the likelihood of this failure mode requires development of margin protocols that are
414
appropriate to each motion management scheme that implies careful commissioning of the
415
motion management system and communication with and training for the physicians who
416
prescribe its use. Absence of clinical practice guidelines related to uncertainty management
417
QA is a national problem. Relatively few institutions measure geometric uncertainties for
418
their patient populations and guidance on performing and analyzing these measurements and
419
determining the validity of margin recipes is urgently needed.
420
Rank
#9
RPN
306
Step#
46
Process
Step
6. Initial Tx Specify special instructions (viz. pacemaker,
Planning
allergies, voiding bowel prep, etc)
Directive
FM: Special instructions not given. Wrong special instruction (e.g. allergy,
pacemaker)
421
422
Special instructions errors take two forms: (i) failure to communicate clinical site- and patient-
423
specific special requests that are required to ensure optimal clinical outcomes and (ii) failure to
424
screen the patient for general contraindications to RT, e.g., pregnancy, or to special conditions
425
required for simulation, other imaging procedures or treatment. The severity of such problems
426
ranges from catastrophic (allergic reaction or unknown pregnancy) to less than optimal imaging
427
(missing contrast). Patient specific special request errors include failure to extract decayed teeth
428
from head and neck patients prior to RT or neglecting to specify appropriate simulation
429
instructions (contrast, omitting bowel/bladder voiding etc). Incidence of type (ii) failures can be
430
reduced through careful screening of all patients using physician consultation checklists and
431
check-off boxes requiring active physician response in a consultation report generator based on
432
the site-specific protocol (Example Checklist 1 in Table IV). Downstream in the process
15
433
(simulation, treatment planning, treatment), personnel should have to acknowledge actively, for
434
example, by signing, specific requests of the patient’s treatment protocol.
435
Rank
#10
RPN
303
Step#
127
Process
Step
8. Treatment 14. Evaluate plan (DVH, isodose, dose tables,
Planning
etc)
FM: 1. Inadequate evaluation
436
437
Errors such as inadequate evaluation are highly ranked because they lead to systematic use of an
438
inferior plan. Unless this is identified at the time of occurrence, it is unlikely to be detected later
439
in the delivery process. Among the progenitor causes are lack of training, poor communication,
440
and lack of attention, though lack of attention is less likely for treatment planning because the
441
planner is directly involved interactively with the planning process and other persons (i.e., the
442
physician) must approve the plan before use. Rushing, a common potential cause of failure, is
443
often due to inadequate staffing or insufficient time allowed for planning.
444
considerations prevent the obvious solutions to this problem - hiring more dosimetrists or
445
planners, or scheduling more planning time per patient. Notwithstanding the economics,
446
departments and professional organizations should persist in making a strong case for adequate
447
staffing. While an ideal QC approach would intercept and correct errors from all these human-
448
factor causes, addressing each cause individually can expend a great many resources.
Economic
449
450
Evaluation failures can also result from selection of inappropriate evaluation parameters by the
451
user or by software.
452
incomplete isodose lines or DVHs, or using non-standard displays, so the evaluator misinterprets
453
the results. For example, displaying only high isodose levels around target volumes and omitting
454
lower isodose levels or critical structure contours can lead to selection of a plan that allows too
455
much dose to critical structures. Review of lower isodose levels can also help reveal dose
456
contributions of individual beams to the dose distribution, and can highlight suboptimal selection
457
of beam angles. Planning software that permits volumetric display of isodose surfaces or that
458
takes the user to the planes with the highest doses is helpful in evaluating hot spots in
459
‘nonspecific’ normal tissue; failing to do this can also lead to selection of a poor plan. Not
460
displaying DVHs for dose-limiting structures can cause similar failures, and can lead to approval
461
of plans that violate protocol or standard tolerances. Changing common display characteristics
For example, failures include basing plan review on the wrong or
16
462
for isodose lines or DVHs (colors, units such as percentages or absolute dose, axes or scales) can
463
lead to similar misinterpretations.
464
465
QM for these human errors is provided by standardized procedures for which graphic output to
466
display and what display techniques to use. Anything specified by the plan directive and any
467
structure noted in the site-specific protocol should be part of the dose and DVH review process.
468
If this information is standardized, then dosimetrists can provide QC during the planning
469
process, and physicists and others can do QA reviews while preparing the plan for approval. If
470
there is no standardization, then the only review may be the review provided by the MD at plan
471
approval – a single point of failure.
472
473
Software failures related to plan evaluation parameters can be difficult to detect, even with the
474
above QM and systematic QC. Commissioning of all the dose display and evaluation
475
functionality of the TPS is crucial. Dose display checks are straightforward, but DVHs are much
476
more problematic, since they are often sensitive to resolution, grid, technique, and anatomical
477
modeling issues.
478
important if one is to characterize the accuracy of those calculations, an important part of
479
detecting software errors in the DVH capabilities. Detecting software failures involves detailed
480
comparisons of plan results with rigorously defined test cases, and against similar plan
481
benchmarks that have been extensively investigated. Software manufacturers should provide
482
tools to facilitate testing and benchmarking of individual plan parameters and evaluation metrics
483
against standard results19-22. The reports from TG5318 and the IAEA TRS-43023 give a great deal
484
of guidance for general commissioning, and quality assurance for treatment planning systems.
Extensive commissioning of DVH calculation and display methods is
485
486
The most practical approach to intercept the human-factor related failure modes is a QA review
487
of the treatment plan after completion. This review must check both qualitative and quantitative
488
aspects of the plan, and the physician, physicist and dosimetrist/planner should all be involved.
489
Since the physician evaluation and approval of the plan was defined as the subject of step 9 in
490
the model process, here we concentrate on the more technical aspects of the plan quality and
491
checks that are typically performed by the physicist. Although the specific order and methods of
492
plan validation vary from facility to facility, a formal checklist of issues to be confirmed or
17
493
evaluated will significantly improve the quality of the QA that results from these checks. Guided
494
by a checklist or structured evaluation process, the medical physicist (or a senior dosimetrist)
495
should assess:
496

The use of the appropriate image sets (e.g., the phase in a 4D CT);
497

Contouring and segmentation of anatomy including the consistency of the contouring of
498
OAR across the datasets; definition of beams and IMRT delivery technique, special field
499
adaptations;
500

Target coverage and uniformity;
501

Sparing of the organs at risk
502

Use of appropriate treatment aids and immobilization;
503

Deliverability of the plan, and other issues that may be particular to the facility.
504
505
It is essential that the physicist’s and physician’s evaluation and checks overlap significantly and
506
that there is good communication between the two. This medical physics QA review of the plan
507
addresses most of the treatment planning failure modes, and also acts as a QC review in
508
preparation for the physician’s “Plan Approval” step, preemptively addressing many of the
509
potential failures that may occur at that step. Example Checklist 3 in Table VI recommends items
510
to include in the treatment plan check.
511
Rank
#11
RPN
283
Step#
40
Process
Step
6. Initial Tx Specify images for target and structure
Plan
delineation
Directive
FM: Specify incorrect image set (viz. wrong phase of 4D CT, wrong MR, etc)
512
513
Specifying image sets to be used for target delineation, particularly when they are obtained
514
outside Radiation Oncology, is a serious potential source of difficult-to-detect errors in the
515
planning process. In many centers, this process consists of the attending physician and/or
516
resident reviewing imaging studies available for the patient using the radiology PACS system.
517
The desired image set is then identified and its study number is passed to the
518
dosimetrists/physicist who contacts the appropriate radiologic technologist and requests them to
519
export the desired DICOM dataset into the RTP file server. There are many potential sources of
18
520
error, including propagation of an incorrect ID number, miscommunication (between, physician,
521
dosimetrist/physicist and radiology technologist), and the possibility that the radiology
522
technologist will manually export an incorrect image set. Because of the growing number and
523
variety of MR and PET imaging studies that are used for planning, the dosimetrist and physicist
524
cannot, on their own, verify the correctness of the secondary image sets imported into the
525
planning system. Only if the physician notices that an incorrect dataset has been selected will
526
the error be detected.
527
528
Ways to change the process to decrease the likelihood of this failure mode include:
529
1. Obtain a modern PACS system that allows the physician to directly download the desired
530
studies when they are viewed. This eliminates the opportunity of miscommunication.
531
2. Expand the site-specific protocol to include the technique factors (e.g., MR pulse sequence,
532
contrast, patient position, volume) to be used for each major clinical site and presentation.
533
This will provide a basis for verifying the image datasets selected for planning.
534
535
536
537
3. Require the physician to complete an online form that not only identifies PACS study ID, but
also the date of the procedure and imaging technique desired.
4. Require the dosimetrist to verify that the imported secondary dataset is consistent with 2 and
3.
538
539
As illustrated by the FTA, a QC check placed prior to importing DICOM images from the PACS
540
server (where full technique information is available) to the RTP (where full DICOM header
541
information is not reviewable in the model for the facility considered) reduces the probability of
542
error.
543
544
A review of the FTA indicates that many of the initial planning directive error pathways can be
545
managed by the same strategy: dosimetrists perform QC checks of the inputs into the planning
546
process (by comparison against the treatment protocol) while the more comprehensive
547
downstream physics QA check of the plan use the same information for the final QA check. For
548
example, to ensure that use of incorrect secondary-to-primary image registration techniques is a
549
detectable error (step 43, rank 23) requires that the treatment protocol documents the standard
550
registration process to be used for this clinical site (which image set is primary, type of
19
551
registration used, e.g. manual vs. automated, which landmarks to align, etc.) This places the
552
onus on the attending physician to request and document variances from the standard procedure
553
where medically indicated. It should be noted that primary image set selection errors are not
554
limited to cases where additional secondary imaging studies must be imported. Commonly,
555
multiple CT simulation datasets could be available (repeat exams for changing medical condition
556
or to address a simulation error; picking the wrong dataset for an adaptive replan, etc.). Other
557
errors that can be intercepted this way include incorrect specification of goals and constraints
558
(step 47, rank 26) and treatment planning approach/parameters (step 45, rank 84). The treatment
559
protocol can be implemented as a patient-specific form to be inserted in the patient’s chart or in
560
the electronic record. Default or standard choices (e.g., DVH planning constraints) would be
561
printed in the form, so if the physician wants to modify these values, the default number is
562
crossed out and the physician-specified number written by hand or as an override.
563
eliminates transcription errors characteristic of a form with simple blanks. On the other hand, it
564
introduces the potential failure of using the default in error because the physician neglected to
565
make a change.23
This
566
567
Developing a detailed and comprehensive policy on 4D motion management is crucial, including
568
indications for 4D vs. 3D simulation CT, indications for using specific respiration sensors or
569
surrogate breathing motion markers, criteria for gated vs. free breathing treatment, and which
570
images (MIPS, slow CT, breathing phase CT closest to average) are to be used for ITV creation,
571
dose calculation, and for generation of reference digitally reconstructed radiographs (DRR).
572
Without specific policies, there is no way to verify that incorrect methods are not being used.
573
Numerous issues are directly involved in registering 4D images, e.g., which landmarks are to be
574
used by therapists in performing online registration of daily gated radiographs and reference
575
DRRs, the accuracy that is to be expected or demanded, and what to do when expectations are
576
not achievable. All these should be addressed in this protocol. An explicit check of the
577
registration used for RTP anatomy definition is an important QA step, and is mentioned in
578
Example Checklist 2 in Table V.
579
Continuing to look at the process steps dealing with motion management, review of the FTA and
580
FMEA potential failure rank 13, step 78, “Inability to support 4D data” reveals four different
581
error scenarios for failures involving motion compensation:
20
582
583
584
585
586
1. The physician specifies an incorrect approach to uncertainty management (e.g., failing to
order intra-fractional imaging for a frameless SRS treatment).
2. The motion management protocol is correctly selected, but planning specifications (e.g., PTV
margin) are inconsistent with protocol
3. The physician correctly specifies the motion management technique and consistently
587
specifies other planning or treatment directions.
However, later physics, dosimetry, or
588
therapist actions are not consistent with policies underlying the physician’s directive (e.g. the
589
wrong CT image set is used to generate the reference DRRs)
590
4. Motion management, all associated planning directives and all subsequent technical actions
591
are consistent with procedures but the motion management technique is inadequate compared
592
to the actual geometric uncertainty characteristic of the patient or relevant population of
593
patients.
594
595
Intercepting errors arising from scenarios 1–3 can be accomplished by written procedures as part
596
of those described above, which clearly identify indications for gated treatment including
597
immobilization, setup, intra-fraction motion monitoring, and planning procedures. The treatment
598
protocol allows the dosimetrist to perform QC on the inputs to treatment planning and
599
subsequent steps. This check should detect variances from established policies and provide a
600
mechanism for negotiating either compliance or a documented variance with the attending
601
physician. The Task Group also recommends incorporating review of motion and uncertainty
602
management techniques into the physicist review of treatment plans.
603
604
Scenario (4) arises not from a random procedural error or mistake but systematic errors due to
605
inadequate commissioning of the motion management process. Reducing incidence of motion
606
management failures is discussed in relation to the step 205 (rank 8).
607
Rank
#12
RPN
282
Step#
106
Process
Step
8. Tx
Calculate dose to optimization points, and final
planning
dose distribution
FM: Really bad beam modeling
608
21
609
Addressing this failure mode requires careful commissioning and documentation of the methods
610
to be used clinically, and, with planning systems that implement several different calculation
611
methods, a check after planning to confirm that the correct dose calculations methods were used.
612
Good applications training by the vendors as well as continuing education by professional
613
societies or in-house in-services can also make physicists and dosimetrists aware of the
614
capabilities of their planning system and aid them in conveying such information to the
615
physicians. The report of Task Group 5318 provides guidance for commissioning beam models
616
and dose calculation algorithms.
617
Rank
#13
RPN
278
Step#
78
Process
Step
7. RTP
4-D representations
Anatomy
FM: Inability to support 4-D data
618
619
The inability to support 4-D data within the planning process should be discovered and
620
characterized during the commissioning process for the planning system and for the treatment
621
planning process involving clinical sites in which respiratory motion (for example) is an
622
important factor. Further handling of motion issues is described above (Rank 8 Step 205, and
623
Rank 11 Step 40).
624
Rank
#14
RPN
278
Step#
44
Process
Step
6. Initial Plan Motion and uncertainty management
Directive
(includes PTV and PRV)
FM: Specify wrong motion-compensated Tx protocol, specified margin size
inconsistent with motion management technique, specified duty cycle and
breathing phase inconsistent with margin for gating
625
626
This FM was addressed with Failure ranks #8 and #11.
627
Rank
#15
RPN
273
Step#
168
Process
Step
10. Plan
9. Prepare e-chart
Preparation
FM:1. Incorrect Tx info. 2. Wrong Rx. 3. Wrong patient/plan
628
629
This step entails transferring the treatment plan information from the treatment planning system
630
into the treatment delivery system. This process may be straightforward or may require many
22
631
intermediate steps. The most error-prone situation involves manual entry into the treatment
632
delivery system computer. The likelihood of errors in programming or interfaces between
633
systems increases with the number of steps and systems involved, so the QM required also
634
increases. Regardless of the process, a QA verification of the data in the treatment unit computer
635
is needed for every patient plan loaded into the delivery system, as any error that occurs during
636
this process will very likely lead to a systematic error in treatment delivery24. Verification at the
637
first treatment is the main subject of Example Checklist 4 in Table VII.
638
Rank
#16
RPN
272
Step#
57
Process
Step
7. RTP
Dataset registration (fusion) of multiple datasets
Anatomy
FM: Poor/wrong fusion (mispositioning, mis-orientation, distortion)
639
640
Issues for dataset registration and fusion of multiple imaging datasets are similar to those
641
described at Rank 11 (Step 40). It is crucial to confirm the accuracy of the registration, as the
642
correctness of any target or other structure based on information in that dataset depends on that
643
registration; such a check is an item in Example Checklist 2 in Table V.
644
Rank
#17
RPN
266
Step#
41
Process
Step
6. Initial Plan Specify protocol for delineating target and
Directive
structure
FM: Incomplete/incorrect list of specified structures and corresponding image
sets, or CTV incorrectly contoured
645
646
Some institutions use specific protocols describing the way various structures are contoured or
647
defined and this protocol decision can be defined separately from the act of contouring (a later
648
step in the process). Most institutions have an internal contouring protocol that describes the
649
typical division of labor (e.g. physicians contour GTV and CTV, dosimetrists’ contour most
650
normal tissue structures). Since five groups (attending physicians, radiation oncology residents,
651
dosimetrists, physicists and physics residents) may perform or review contours at different times,
652
the potential for miscommunication and misunderstanding is high. Standardizing procedures
653
provides a basis for challenging physician segmentations and for reducing the guesswork and
654
decision-making by dosimetrists on how and what to contour. Thus, it is recommended that
23
655
contouring protocols be included in each site-specific protocol. Contouring protocols should
656
specify the contouring process, from structure naming conventions and display colors to who is
657
responsible for what contours, who checks them, what techniques are used (manual vs.
658
automated vs. mixed), and what Boolean combinations of structures are needed, as well as
659
specifying anatomical guidelines for contour drawing. To the extent possible, written guidelines
660
for physician contouring tasks (e.g., what CT-visible landmarks to use to minimize prostatic
661
apex-, mid gland-, and base-delineation errors25) should be developed and disseminated to the
662
planning group. Physicists and dosimetrists can participate in QC of these segmentations. As
663
the FTA shows, an independent pre-IMRT planning QA check of the image segmentation
664
instructions and the resulting contours can help identify errors. As the annotated FTA (Appendix
665
D) indicates, a single QA check can address a number of major error pathways for the anatomy
666
definition, from step 125 (secondary image set specification) to step 165 (motion/uncertainty
667
management). Once contouring protocols are chosen, at least parts of these checks may be
668
performed by a physician, physicist or dosimetrist.
Rank
#18
RPN Step#
Process
265
94
8. Tx planning
FM: Wrong plan
Step
2. Enter prescription + planning constraints
669
670
In this FM the radiation oncologist and/or treatment planner specifies an incorrect prescription or
671
evaluation constraint for a treatment plan. This is a highly ranked failure because it is difficult to
672
detect through the routine clinical process. Several QM measures may intercept this error. The
673
treatment planner or medical physicist can more easily detect the problem if the institution has
674
clearly defined treatment protocols. Although there are variations between patients, questioning
675
the physician about unusual requests (communication) before planning and/or at the plan review
676
stage would intercept erroneous prescriptions or evaluation guidelines and planners should be
677
empowered to raise such questions. Errors can also arise from the entry of erroneous information
678
into the planning system by the planner. There are two opportunities to detect such errors: the
679
physician’s review of the plan or the physics plan check. In both cases, comparison between the
680
generic site-specific protocol, the initial planning directive and the prescription and evaluation
681
criteria actually used can identify errors. But well-defined guidelines for each treatment protocol
682
are a key component in ensuring that this failure mode does not propagate to treatment.
24
683
684
Related to this failure mode would be erroneous entry of the parameters into the optimizer by a
685
dosimetrist, which can produce a suboptimal plan with a medium severity score. Such failure can
686
be detected if the optimization parameters are checked before optimization during the physics
687
check of the plan (Example Checklist 3 in Table VI) using standardized planning protocols or
688
individualized plan directives, or by either the physicist or the physician if inferior plan quality is
689
noted and questioned.
690
Rank
#19
RPN Step#
Process
Step
265
85
8. Tx planning 1. Specify ROI for optimization process
FM: 3. Inconsistent length (sup/inf) of ROI
691
692
In this FM the superior-inferior lengths of contoured regions-of-interest (ROIs) are not consistent
693
between individual patient datasets because institutional protocols do not exist on how to draw
694
regions of interest for different disease sites.
695
treatment plan review and lead to inferior plans being delivered. This was addressed with failure
696
rank 10, step 127.
This inconsistency can cause difficulties in
697
Rank
#20
RPN FMEA# Process
258
135
9. Plan Approval
FM: Wrong plan
Step
1. Plan OK to go to Treatment
698
699
In this failure mode, a plan is approved but an earlier (and inferior) version is prepared for
700
treatment. This could be prevented by software that only allowed the treatment planning system
701
to download to the treatment-unit computer a plan with the physician’s (electronic) approval. At
702
a minimum, the physician should be required to sign (electronic, written or both) the accepted
703
treatment plan before it is sent to the machine; this assures that the correct plan is clearly
704
identified. Regardless of whether the above-mentioned special software is available, a physics
705
check performed prior to Day 1 treatment should compare the parameters (MU, gantry angles,
706
etc) in the TPS with those in the treatment management system (TMS) and also compare the
707
intensities from the approved plan in the TPS with the intensities that were downloaded to the
708
linac. It is very unlikely that all these would match if a seriously wrong plan had been sent to the
709
record and verification system.
25
Rank
#21
RPN Step#
Process
Step
246
95
8. Tx planning 8. Enter prescription + planning constraints
FM: 2. Incorrect dose prescription + dose limit constraints
710
711
This failure mode is addressed with rank #18 step 94.
712
Rank
#22
RPN
246
Step#
3
Process
Step
1. Patient
Entry of patient data in electronic database
Database Info
or written chart
FM:2. Incorrect or incomplete previous treatment history
713
714
This is one of the three failure modes described for Process 1, “Patient Database Information”.
715
All involve failures entering data into the patient record (electronic or paper) and all can be
716
addressed at the same time. There are repeated opportunities for physicists, physicians and
717
dosimetrists to check selected parts of the data base in downstream processes (e.g. treatment
718
planning), but this is a highly ranked failure because information about previous treatment may
719
be used in treatment planning and this and other information may be used in other ways in
720
multiple later steps; incorrect or incomplete data compromises these uses. QM for the initial
721
patient database information must be tied carefully to the input of this data and depends in detail
722
on the processes and systems used at individual institutions. The QM should be designed by an
723
interdisciplinary team including physicians and administrators.
724
Rank
#23
RPN
246
Step#
43
Process
Step
6. Initial Plan Specify image registration goals – 4D
Directive
FM: Specify inappropriate protocol or tolerances for registration, or wrong
reference image dataset (primary)
725
726
This FM was addressed with that for rank #11.
727
Rank
#24
RPN Step#
Process
Step
240
189
11. Day 1 Tx Set Treatment Parameters
FM: Wrong Tx accessories (missing/incorrect bolus, blocks)
728
729
See text of main article, section 3, rank 24.
730
26
Rank
#25
RPN
240
Step#
52
Process
Step
7. RTP
Import images into RTP database
Anatomy
FM: 2. Wrong imaging study (correct patient) viz. wrong phase of 4D CT
selected for planning, wrong MR…
731
732
This FM was addressed with that for rank #11, step 40 and is part of the pre-planning anatomy
733
check.
734
Rank
#26
RPN
234
Step#
47
Process
Step
6. Initial
Specify dose limits, goals and fractionation
Plan
Directive
FM: Inappropriate or incomplete target doses and/or normal tissue constraints
specified or assumed
735
736
As with other FMs in the initial plan directive, (Rank 11 step 40, Rank 14 step 44), errors must
737
be intercepted early in the process to prevent them from affecting the entire treatment. The
738
incidence of this particular FM can be greatly reduced by use of standard protocols for site-
739
specific clinical treatment, peer review and physics review (both referring to the protocol).
740
Rank
#27
RPN
234
Step#
187
Process
11. Day 1
Tx
FM: 2. Images misinterpreted
Step
Localization (portal images and/or other
localization devices)
741
742
This FM has been addressed in the discussion for rank #11, step 40.
743
Rank
#28
RPN
231
Step#
128
Process
Step
8 Treatment 15. Evaluate plan (DVH, isodose, dose tables,
planning
etc)
FM: Incorrect DVHs, IDLS, etc
744
745
This failure mode is generally addressed with FM rank #10, step 127.
746
Rank
#29
RPN
230
Step#
66
Process
7. RTP
Anatomy
Step
Boolean combination of delineated structures
27
FM: 1. Wrong structures combined
747
748
This failure mode is addressed with FM rank #2 step 58.
749
Rank
#30
RPN
230
Step#
108
Process
Step
8. Tx
Heterogeneity correction
planning
FM: 1. Wrong or poor algorithm
750
751
Another class of systematic planning failures is illustrated by considering a planner who
752
repeatedly invokes the wrong dose-calculation algorithm (Pencil Beam instead of
753
Superposition/Convolution) or uses a suboptimal leaf-sequencer choice on all patients because
754
the appropriate procedure is not defined, followed or understood. Prevention of this type of error
755
requires documentation of the methods to be used clinically, software which documents the
756
algorithms used in a specific plan, planner education and training on the appropriateness and
757
limitations of the use of available dose calculation algorithms for various disease sites, and a
758
check after planning which confirms that the correct methods are used.
759
Rank
#31
RPN
229
Step#
107
Process
8. Tx
planning
FM: 2. Poor beam modeling
Step
Calculate the dose to optimization points and
final dose calculation
760
761
Measuring necessary beam data, entering it into the treatment planning system, parameterization
762
and validation of the dose calculation algorithm all constitute high-risk procedures, since errors
763
in any of these steps can result in systematic dose calculation errors affecting many patient
764
treatments. Failures may result from poor data collection, poor beam modeling or inadequate
765
verification during the treatment planning system commissioning. Data collection and beam
766
modeling for IMRT treatment planning systems requires experience and thorough understanding
767
of the scanning system, data required by the planning system, and beam modeling process. The
768
manufacturers of certain treatment machines or planning-system dose-calculation models
769
commission the systems. Whether performed by the institution’s physics staff or the vendor,
770
after beam modeling, a comprehensive set of verification tests should be performed to ensure
28
771
that the system calculates the dose distributions correctly. Details for commissioning of IMRT
772
can be found in AAPM IMRT subcommittee Reports.26,27
773
774
A poorly implemented, inadequate, or incorrectly parameterized heterogeneity-correction
775
algorithm, or a limited planning system that does not provide the best algorithm for all cases,
776
compromises the quality of the treatment. Commissioning of the calculation algorithm should be
777
one of the key components of treatment-planning system commissioning.18,28 For those who
778
have access to another planning system, one way to detect issues with a new algorithm is to
779
recalculate patient dose distributions with a second, independent dose calculation system that had
780
previously been
781
[http://www.aapm.org/pubs/reports/RPT_119.pdf]27 for evaluating IMRT planning systems
782
together with the results obtained when the TG-119 members ran these tests on their own
783
systems. Evaluation with anthropomorphic phantoms from the RPC is required for institutions
784
participating in nationally funded radiation therapy protocols. Detailed review of the calculated
785
dose distributions, particularly for test cases that can delineate the behavior of the algorithm or
786
for which accurate knowledge of the expected results exists, can be performed by the medical
787
physicist.
788
include robust and comprehensive commissioning and, potentially, a QC process with an
789
independent volumetric dose calculation system capable of heterogeneity corrected calculations.
790
Inadequate software usually cannot be addressed quickly, but at least those involved in the
791
treatment planning process can understand those situations for which the algorithm performs
792
poorly. An informed physicist can educate physicians and administrators and encourage purchase
793
of improved software in a future budget cycle.
commissioned.
Task
Group 119 has
provided a suite of tests
Assuring correct dose calculation for heterogeneity corrected IMRT plans must
794
795
In general, the treatment-unit input data for the planning system must be validated. Part of the
796
validation process could be an independent review by a second medical physicist, while another
797
part would be the in-phantom testing discussed in the previous paragraph. Although there is
798
little danger of the treatment planning software spontaneously becoming corrupt, modern
799
planning systems are complex software systems with millions of lines of code, multiple
800
computers, processes, databases, data files, and complex interfaces with other systems, so there
801
are many possible problems that can occur as any of those components are modified or updated,
29
802
including issues like virus-checking software and operating system patches. Hazards often
803
appear after updates to the computer system hardware or software, or to other computer systems
804
on which treatment planning relies, such as a CT unit. Such updates may be made without the
805
knowledge of the responsible physicist, particularly with respect to imaging systems in other
806
departments. Some mechanism should be implemented for monitoring such service and for
807
verifying that these services or upgrades, updates have not corrupted the planning system
808
operation. The report of AAPM Task Group 53 provides guidance on quality management for
809
treatment planning computers.i
810
Rank
#32
RPN
229
Step#
207
Process
Step
12. Day N
Tx machine and peripheral hardware setup for
Tx
Tx
FM: Changed prescription dose (and MU) occurring after initial Tx and not
entered into chart and/or treatment unit computer
811
812
See text in main article, section 3, Rank 32. In addition to this FM, there are a number of other
813
related FMs for Day N Tx that have high RPN numbers (Rank 34 step 208, Rank 40 step 202,
814
Rank 42 step 206, Rank 63 step 204, Rank 152 step 203).
815
Rank
#33
RPN
228
Step#
80
Process
Step
8. Tx
Specify ROI for optimization process
planning
FM: 1. Incorrect classification of a structure as overlapping or non-overlapping
816
817
Handling of overlap regions between target volumes and critical structures differs significantly
818
among treatment planning systems.
819
dependent on this step and users must ensure that the overlap regions are handled correctly and
820
are consistent with the optimizer requirements. This FM is generally best addressed through
821
both training and QC measures. Commissioning should assure that physicists and dosimetrists
822
understand how the TPS and its optimizer handle the overlap region. Some older TPSs had
823
limitations on the number and nature of allowed overlapping structures though this issue has
824
largely been resolved in more recent versions. If the TPS is affected by these kinds of limitations
825
it is critical that the medical physics check of the plan (Example Checklist 3 in Table VI)
826
includes a confirmation of the correct choices for the relevant structures.
The quality of a treatment plan is often significantly
30
827
Rank
#34
RPN
228
Step#
208
Process
Step
12. Day N
Adapt to changes
Tx
FM: Changes in prescription dose (hence in MU) occurring after initial Tx not
entered into chart and/or R/V
828
829
This is essentially the same failure as discussed in Sec. 9 of the main body of the TG 100
830
protocol relating to the Rank 32 FM, step 207.
831
Rank
#35
RPN
226
Step#
199
Process
Step
12. Day N
Patient setup for treatment
Tx
FM: Special patient prep (e.g., full bladder) not done
832
833
Incorrect preparation of the patient (bladder filling, bolus placement, specific immobilization
834
aids) may occur for a single fraction, with minor severity or many treatments with potentially
835
serious impact on the treatment outcome. A standard QM method includes weekly physics and
836
therapist chart reviews in order to limit the impact of “Day N” failures (see Example Checklist 5
837
in Table VIII). The concern is lower on the first day of treatment because all the staff is looking
838
at the set-up carefully, but over time, the likelihood of missing patient-specific orders increases.
839
Organizational factors, such as suboptimal staffing, that lead to rushing and fatigue, will increase
840
the frequency of such failures, though these can be partly mitigated by managerial actions.
841
Training and policies and procedures are also important (as for all other steps in the process).
842
Requiring the therapist to acknowledge specific preparations in the treatment-unit computer or
843
paper chart before treatment begins helps reduce the frequency of such events, but without
844
interlocks or similar automated preventive mechanisms, reminders cannot eliminate these
845
omissions completely. Rare, sporadic failures will not likely affect outcomes. However even
846
sporadic failures should not be taken complacently, and occurrences should be used as
847
opportunities to discover and correct their causes.
848
Rank
#36
RPN
225
Step#
139
Process
Step
9. Plan
2. Completion of formal prescription after
Approval
planning
FM: 2. Wrong total dose, fractionation
31
849
850
In this FM, the formal prescription adopted when the plan is approved is not what the physician
851
intended. Since this is a physician decision, there are few physics QM steps possible. The first
852
is the use of standardized protocols, and a department requirement that the plan directive be clear
853
before the planning process begins, so that a change after planning becomes rare and very
854
noticeable. Another possible QM involves peer review of the plan approval decision, including
855
all the parameters involved (including the dose and fractionation). There should be careful peer
856
review of this information by the end of the first week of treatment – or sooner for few fraction
857
treatments. Furthermore, any change in the dose prescription during treatment should be
858
documented and communicated to the appropriate personnel.
859
Rank
#37
RPN Step#
Process
Step
224
190
11. Day 1 Tx Treatment patient – monitor treatment
FM:1. Failure to notice patient move
860
861
The majority of patients are treated without formal motion management methods and for them
862
there are currently very few routine ways to monitor intra-fraction motion. If the therapist does
863
not observe this and act on it, the error goes undetected and uncorrected. For these patients, the
864
sole QM method is careful observation during treatment, so delivery is stopped if there is
865
significant motion.
866
867
In recent years, a number of commercial systems that monitor patient position during treatment
868
(e.g. video systems, RF transducers) have become available. These systems could be designed to
869
pause the beam or trigger an alarm if the target moves outside a user-set tolerance set and could
870
address the need for extreme human attention. These systems must be commissioned with
871
protocols and action levels defined, and QA for the devices themselves added to the process.
872
Rank
#38
RPN
224
Step#
42
Process
Step
6. Initial
Specify image registration goals – MR/PET
Plan
Directive
FM: Specify inappropriate protocol or tolerances of registration or wrong
reference image set (primary)
873
32
874
This failure mode is addressed in the Rank 11 step 40 and Rank 23 step 43 discussions, and
875
relevant checks are included in the anatomy checks prior to treatment planning.
876
Rank
#39
RPN
223
Step#
173
Process
10. Plan
Prep
FM: Incorrect modification
Step
11. Manual plan modification
877
878
This FM is the highest ranked error in the plan preparation step because almost any error can be
879
made during manual modification of the plan, making potential severity high. All electronic
880
systems must allow information to be edited in order to correct errors caused by other things or
881
to make delivery smoother. Therefore, the QM for this FM is dependent on process control and
882
human checks. By the time the plan preparation step is reached, the treatment plan has been
883
completed and approved for clinical use; the next step is Day 1 Treatment. Therefore, two types
884
of QM are possible: 1) control, documentation and checks of any manual plan modification
885
during plan preparation, and 2) a routine check at, or shortly before, Day 1 that all information is
886
still correct. To control this FM, both of these approaches should be followed. Any plan
887
changes made during the preparation step should force a check by another individual to confirm
888
that 1) the change was appropriate and 2) that change was carried out and documented correctly.
889
Secondly, the Day 1 check should confirm accuracy of the plan information, including
890
prescription, fractionation, MU, technique, and all other critical parameters (see Example
891
Checklist 4 in Table VII).
892
Rank
#40
RPN
223
FMEA#
202
Process
Step
12. Day N
Patient setup for treatment
Tx
FM: Info in R/V system improperly changed
893
894
See discussion in rank 32 step 207 in the main body of the report. Modern delivery systems
895
allow locking the treatment delivery information to prevent unauthorized changes, and this
896
feature should be employed to help reduce accidental errors. Note that it does not prevent
897
intentional (but incorrect) changes made by authorized individuals.
898
Rank
RPN
Step#
Process
Step
33
#41
222
39
5. Transfer images
Handling of DICOM_RT objects
and other DICOM
(other than images) between the
data
scanner and the planning system
FM: Incorrect transfer of interest points or contours
899
900
Image information forms the basis for most treatment planning, so correct transfer of points and
901
contour information from images or from the CT-Simulation software is crucial. QM for this
902
issue includes careful commissioning of the transfer of any information from other systems (e.g.,
903
CT) to the planning system. However, commissioning is not enough, as there is a very large
904
range of information that may be transferred. Therefore, QM for this FM also requires that each
905
patient’s imported contours and points of interest be visually inspected and evaluated prior to
906
treatment planning. Some contour import discrepancies can be demonstrated by misalignment
907
between overlaid contours and underlying image anatomy, for example. Incorrect transfer of
908
interest points is much more difficult to detect, and requires comparison of the points against
909
anatomical locations or fiducial markers or coordinates on the relevant imaging study. This is
910
especially important for interest points which indicate planned beam isocenters or correlate with
911
marks on the patient’s skin. If such interest points are mis-located during image transfer, there is
912
a possibility that this may translate to a treatment to the wrong anatomy or site. Transferred
913
isocenter locations should be verified against skin wires (markers), bony landmarks or implanted
914
fiducials where possible. In addition, patient setup photos that were acquired during the
915
simulation scan can be useful for major misalignments.
916
Rank
#42
RPN
222
Step#
206
Process
Step
12. Day N
Tx machine and peripheral hardware setup
Tx
for treatment
FM: Changes in # of fractions occurring after initial treatment (increase or
decrease) incorrectly scheduled or not applied to patient’s Tx
917
918
See text in Sec. 9 in the main body of the TG 100 protocol relating to the Rank 32 FM.
919
Rank
#43
RPN Step#
Process
Step
222
191
11. Day 1 Tx Tx patient – monitor Tx
FM: 2. Failure to notice inappropriate machine operation
920
34
921
Failure to monitor machine operation during treatment is a crucial error, as it may (in very
922
unlikely situations) lead to catastrophic harm to the patient (e.g., the Therac 25 accidents29 or
923
recent IMRT errors30). At present, an alert therapist is an important guardian against such failures
924
and the relationship between therapist and physicist must encourage the reporting and
925
investigation of odd machine behavior. As discussed in the main document under “Summary,
926
Treatment Machine QM Design”, reliance solely on an attentive human is suboptimal and
927
manufacturers should prioritize development of interlocks to prevent hazardous events during
928
treatment.
929
Rank
#44
RPN
221
Step#
83
Process
8. Tx
planning
FM: 2. Incorrect ROI volumes
Step
Specify ROI for optimization process
930
931
QM for this issue is virtually the same as that for rank 19 step 85, and is addressed with the QM
932
for FM rank 10, step 127.
933
Rank
#45
RPN
220
Step#
93
Process
Step
8. Tx
Enter prescription + planning constraints
planning
FM: 1. Incomplete or incorrect set of objectives and constraints
934
935
This FM is the result of poor communication between the radiation oncologist and the treatment
936
planner, or human error in entering the information that has been communicated already.
937
Inadequately defined treatment planning procedures and poor communication can result in
938
delayed planning and substandard treatment plans. This FM can be prevented most effectively
939
by site-specific planning and evaluation protocols and well-designed communication methods
940
between the radiation oncologist and treatment planners. Physics check of the treatment plan
941
should review the parameters used.
942
Rank
#46
RPN
219
Step#
68
Process
Step
7. RTP
Boolean combination of delineated structures
Anatomy
FM: 3. Resulting structures ambiguously or incorrectly named resulting in
incorrect correspondence of Tx goals/constraints and structures
35
943
944
This FM is discussed in rank 2 step 58, where standardized protocols and site-specific anatomy
945
checks were recommended to prevent this failure.
946
Rank
#47
RPN
216
Step#
96
Process
Step
8. Tx
2. Enter prescription + planning constraints
planning
FM: 2. Incorrect dose prescription + dose limit constraints
947
948
This FM results from a planner entering an incorrect dose prescription, dose constraints, or
949
unrealistic planning objectives. This could be due to human error or because the planner is not
950
knowledgeable on how the optimization algorithm works.
951
minimize a score function and ignore possible inconsistencies between planning objectives and
952
constraints. Not having the logic check for conflicts in optimization objectives is a major
953
limitation of many systems. Optimizing on an incomplete or incorrect set of objectives and
954
constraints can result in long optimization times and/or lack of convergence, and eventually will
955
result in substandard treatment plans. When a physician is presented with a plan that was created
956
with inadequate optimization parameters, it may be very difficult to differentiate between poor
957
plans due to optimization parameters or simply due to challenging tumor and surrounding critical
958
structure anatomy. Review by an experienced physicist or dosimetrist may identify inferior plans
959
that can be improved (at the cost of more planner time). Site-specific planning protocols (e.g.
960
beam angles, generic constraints that often work) and QC steps to verify consistency and
961
completeness of planning objectives through QC steps should reduce these problems. Software
962
manufacturers should provide tools to facilitate benchmarking and verification of individual plan
963
parameters and evaluation metrics against standard results. Additionally, thorough training and
964
regarding the planning system and optimizer functionality should be provided to all involved in
965
the planning process. Finally, researchers and planning system vendors should continue to
966
explore and make available more efficient optimization algorithms.
Most modern IMRT optimizers
967
Rank
#48
RPN
216
Step#
64
Process
Step
7. RTP
PTV Construction
Anatomy
FM: Use margin width or protocols for PTV that are inconsistent with dept.
procedures
36
968
969
Mitigation for this FM has already been discussed above with rank 6 step 65.
970
Rank
#49
RPN
213
Step#
147
Process
Step
10. Plan
1. Entry of demographic info
Prep
FM: 3. Critical patient info not recorded
971
972
A careful QA check of patient information regarding such factors as previous treatment and
973
presence of radiation-sensitive implanted devices should be performed by the physician prior to
974
plan preparation. Planners and therapists should be encouraged to raise questions if such
975
information seems inconsistent.
976
Rank
#50
RPN
213
Step#
159
Process
Step
10. Plan
4. Prepare DRRs and other localization
Prep
imaging data
FM: Incorrect images (wrong angle, divergence, etc).
977
978
There are many factors that can cause incorrect images to be prepared as DRRs or other relevant
979
localization imaging, including lack of training or procedures, human failures, and software
980
errors. To prevent the propagation of such errors to treatment delivery, QA during the plan
981
preparation is necessary to confirm that localization images are correct. DRR creation in new
982
systems should be commissioned to avoid software-related errors in this process.
983
Rank
#51
RPN
212
Step#
110
Process
Step
8. Tx
7. Setup optimization parameters and
planning
constraints
FM: 1. Wrong parameters selected. 2. Bad parameters selected.
984
985
This FM is closely related to rank 47 step 96, except that in this case the planner incorrectly
986
selects treatment plan parameters other than the intended dose constraints. These include, for
987
example, beamlet size, step size, algorithm, stopping criteria, maximum number of segments,
988
number of iterations, smoothness of fluence, fixed jaw setting etc. These parameters can be
989
chosen in error or due to poor understanding of the planning system. Corrective actions include
990
training for the planning system, well-developed planning guidelines and procedures, and QC
37
991
measures. QC measures should include an independent review of treatment plan parameters to
992
ensure that they agree with departmental policies and procedures. In some TPS systems, a user-
993
designed script can be created to check the validity and consistency of treatment plan parameters.
994
The scripts are continually run during the planning process, reducing the time wasted due to
995
planning with inadequate parameters and also reducing the errors that are propagated through the
996
planning process.
997
998
Several treatment planning failure modes involve hardware and software design and operation.
999
These problems should be identified during commissioning, where the limitations of the system
1000
need to be established so that operation will remain within those limitations. Any treatment
1001
planning system requires evaluation of the calculation algorithms, data input and handling
1002
mechanisms and output modes. Treatment planning for IMRT requires particular focus on the
1003
operation of the computations for small field segments, leaf sequencing and machine-specific
1004
issues such as field partitioning for large beams or interdigitation limits on the MLC. For each
1005
clinical site or IMRT protocol, the inverse-planning software should be tested for a variety of
1006
anatomical presentations with a wide range of reasonable optimization cost functions and
1007
constraints prior to its clinical use. In-phantom verification of the IMRT plans combines the
1008
testing of the treatment planning system and the treatment unit’s ability to deliver the planned
1009
treatment and should be part of planning system commissioning and periodic QA. Ideally,
1010
constraints and situations should be varied until the optimizer fails to obtain a reasonable plan for
1011
clinical cases, so reasonable limits on optimization parameters are identified. Cases should be
1012
repeated to assess the reproducibility of the program. If this is not achievable then
1013
communication with others who use the same TPS (user groups) may provide information in
1014
understanding the limits of reasonable operation of particular optimization algorithms.
1015
Commissioning should also determine the robustness of the system to operator errors or failure
1016
to follow the expected order of input.
1017
Rank
#52
RPN
212
Step#
197
Process
Step
12. Day N
Patient setup for Tx
Tx
FM: Bolus not applied as prescribed (omitted, wrong position or thickness)
1018
38
1019
QM for this FM is similar to rank 35 step 199, in that this is an error that can happen on any
1020
treatment day, as well as consistently through the course of treatment. Since placement of bolus
1021
is typically an individual patient-related issue, standardization and training often have little role
1022
in preventing this problem. Some treatment management systems require active documentation
1023
of bolus placement and bar-coding may be available. But until such systems are widely available,
1024
the main QC is good documentation in the chart and careful daily review of the specific choice
1025
of bolus and its location. Check lists for set-ups may also be useful requiring one therapist to
1026
place the bolus and a second one to check the box on the list.
1027
Rank
#53
RPN
211
Step#
60
Process
Step
7. RTP
Delineate GTV/CTV and other structures for
anatomy
planning and optimization
FM: 3. Poorly drawn contours (spikes, sloppy, etc)
1028
1029
This FM is similar to rank 2 step 58 and rank 5 step 59, and QM discussed for those FMs will
1030
also deal with this error.
1031
Rank
#54
RPN
211
Step#
120
Process
Step
8. Treatment 10. Run leaf sequencing to create deliverable
planning
plan
FM: 4. Wrong user parameter selection
1032
1033
For some optimization systems or strategies, the plan optimization step is different from leaf
1034
sequencing, in which the intensity distribution obtained by the optimization system is converted
1035
(by a separate optimization algorithm) into a set of MLC trajectories (dMLC delivery) or
1036
segments (sMLC delivery). The process of leaf sequencing can degrade the quality of the final
1037
plan, as compromises are made while converting the unconstrained intensity distributions into a
1038
finite deliverable set of MLC patterns. If the leaf sequencing step has user-defined parameters
1039
(e.g., the number of allowed intensity levels or segments), these can be chosen incorrectly or
1040
poorly. Training and standard protocols can help minimize the frequency of the problem. A QA
1041
check during the physics plan check can then verify that the correct decisions were made.
1042
1043
Just checking that the parameters used in the sequencing are correct does not ensure that the
1044
actual leaf sequences obtained correctly deliver the desired fluence distributions. To verify this
39
1045
behavior, two major activities are required. First, detailed commissioning of the IMRT planning
1046
and delivery system is necessary, including many tests of optimized intensity distributions being
1047
converted to MLC sequences, and then delivered to a phantom, so the measured dose distribution
1048
can be used to confirm that the entire process is working correctly. Testing all fluence patterns to
1049
confirm their correct behavior and resulting intensity distributions before any clinical use is
1050
crucial.
1051
delivery when accompanied by strict QA of the generic performance of the MLC. Both the check
1052
program and the QA procedures should undergo commissioning to confirm that they would
1053
detect aberrant MLC patterns. We also note that pre-treatment verification does not address
1054
potential “Day N” delivery problems. Checking the leaf pattern daily could provide protection
1055
from the most serious effects on the patient, but for a facility with even a third of the patients
1056
under IMRT treatments at any given time, this check would consume a great deal of time and be
1057
clinically unfeasible. The TG urges vendors to provide automated tools to avoid this daily
1058
problem (e.g., use of checksums or other automated checks or the Dynalog files or equivalent)
1059
that might demonstrate the MLC descriptions were identical and unchanged day to day. Similar
1060
automated checks for all files related to a patient’s treatment are also necessary.
Calculational checks of the fluence patterns can be substituted for a physical plan
1061
Rank
#55
RPN
210
Step#
62
Process
Step
7. RTP
Delineate GTV/CTV, etc
Anatomy
FM: 5. Misleading or erroneous structure names and/or other labels (e.g. color)
resulting in incorrect correspondence of treatment goals/constraints and
structures
1062
1063
This FM is similar to rank 2 step 58, rank 5 step 59, and rank 53 step 60, and QM discussed for
1064
those FMs will also satisfy this potential failure.
1065
Rank
#56
RPN
210
Step#
198
Process
Step
12. Day N
Patient setup for Treatment
Tx
FM: Changes in patient geometry (thickness changes due to wt. gain/loss, tumor
growth/shrinkage) ignored or not recognized
1066
1067
This FM is similar to rank 35 step 199, special patient preparation (e.g. bowel preparation) not
1068
performed. Here, however, changes in patient geometry that may affect setup accuracy are not
40
1069
recognized. The most common flag for this situation is that the details of the setup parameters
1070
(couch positions, SSD, etc.) change over time. The therapist is the first line of defense against
1071
this FM. It may also be noticed in weekly physics or therapist reviews by investigation of table
1072
position overrides (Example Checklist 5 in Table VIII). Changes, such as weight loss or tumor
1073
regression, can affect the set-up and dose delivery accuracy. The common practice of weighing
1074
the patient weekly can detect weight changes. Direct observation by the therapist and weekly
1075
physical examination by the physician can detect less obvious changes such as swelling, skin
1076
changes, or increased coughing. For systems equipped with onboard imaging, weekly imaging
1077
to assess the need for replanning may be an effective way to limit deleterious effects of internal
1078
changes in certain patients (this is a field of active study).
1079
Rank
#57
RPN
206
Step#
73
Process
7. RTP
anatomy
FM: Incorrect editing
Step
Editing masks, table, other non-patient data
included in scan info
1080
1081
As with many other parts of the anatomy definition process, standard protocols, training, and an
1082
independent review at the end of the anatomy definition phase, as discussed in relation to the
1083
rank 2 FM can help intercept this error.
1084
Rank
#58
RPN
206
Step#
19
Process
3. CT/Sim
Step
Isocenter and other special point coordinates
recorded in software and/or print
FM: Not recorded or incorrectly recorded (e.g. change in isocenter made but not
documented)
1085
1086
This failure is the first listed for the simulation process. Many failures at this step can propagate
1087
through the entire course of the treatment. Interestingly, the Task Group only identified a few
1088
potential failure causes in the CT simulation process in areas that are avoided by the traditional
1089
physics
1090
(http://www.aapm.org/pubs/reports/RPT_83.pdf).31 This may be due to the difficulty of
1091
imagining a world without standard QA measures or to the inherent robustness of state-of-the-art
1092
CT units. This aspect of the TG100 FMEA should be re-evaluated by readers who use it as a
1093
starting point for design of their own QM program.
QA
measures
that
are
described
41
in
the
report
of
AAPM
TG-66
1094
1095
Incorrect marking or documentation of the isocenter or a reference point determined at the CT, or
1096
changing isocenter after the simulation has happened without accompanying documentation, can
1097
be an extremely difficult failure to identify or correct. As physician time at such a localization
1098
step is often limited, the simulation process needs a QA step at the end where the physician, the
1099
therapist, and all other staff involved review all the crucial information obtained at the procedure
1100
for consistency. This is almost the only opportunity to prevent the use of an incorrect or
1101
inconsistent set of imaging and physical (marks) data. The one other opportunity to detect a
1102
problem here is detection of an inconsistency by the dosimetrist planning the case, by the
1103
physicist doing the final plan check, or by the physician as they review the day 1 setup for the
1104
patient. Use of a simulation procedure form or checklist can help the simulation staff avoid
1105
missing or entering inconsistent information in the patient setup documentation.
1106
Rank
#59
RPN
205
Step#
67
Process
Step
7. RTP
Boolean combination of delineated structures
Anatomy
FM: 2. Wrong Boolean operation(s) used
1107
1108
This FM is discussed in rank 2, step 58, which uses standardized protocols and site-specific
1109
anatomy checks to prevent this failure.
1110
Rank
#60
RPN
203
Step#
181
Process
11. Day 1 Tx or
12 Day N Tx
FM: 1. Incorrect Tx isocenter
Step
Position patient for Tx
1111
1112
This FM describes the isocenter being positioned in the wrong place, due to a device failure.
1113
Other causes of this error include a more subtle error (rank 142, step 182), a human failure
1114
(inadequate training or inattention, rank 77, step 183). This FM can be a crucial one, especially
1115
if the mispositioning happens for multiple fractions or in hypofractionated treatments. One of
1116
the best QC steps available is the daily use of IGRT setup techniques and this is one of the major
1117
benefits of the IGRT concept. Careful IGRT is an effective technique but it should be
1118
accompanied by a departmental protocol limiting the shifts that can be made without a special
1119
investigation e.g., physician inspection and approval. If IGRT is not available, then careful use
42
1120
of the patient setup documentation during the setup procedure is crucial. Imaging to confirm the
1121
correct location as often as possible will minimize the opportunity for this error to continue
1122
through multiple fractions. Other QC can involve using or carefully understanding differences in
1123
coordinates that demonstrate problems with the patient setup – especially if the immobilization
1124
device used for the patient can be registered to the table.
1125
coordinates used for treatment at the weekly physics or therapist chart check (Example Checklist
1126
5 in Table VIII) can also warn the user of problems to be investigated.
In such cases, review of table
1127
Rank
#61
RPN Step#
Process
Step
202
184
11. Day 1 Tx Position patient for Tx
FM: 2. Incorrect patient position
1128
1129
See discussion for rank 60 step 181.
1130
Rank
#62
RPN
200
Step#
169
Process
Step
10. Plan
3. Prepare e-chart
Prep
FM: 1. Incorrect Tx info. 2. Wrong Rx. 3. Wrong patient/plan
1131
1132
See discussion in rank 15 step 168.
1133
Rank
#63
RPN
200
step#
204
Process
Step
12. Day N
Load patient file
Tx
FM:R/V fails to detect incorrect Tx conditions in special situations
1134
1135
See discussion in rank 32, step 207.
1136
1137
This step also contains potentially severe errors that result from mixed hardware
1138
incompatibilities. One example is the possibility of inadvertently having the treatment unit’s
1139
MLC in the field concurrently with an external MLC, such as a micro-multileaf collimator or the
1140
use of incorrect field settings with stereotactic cones. This kind of problem is often associated
1141
with the use of non-standard devices or mixing devices from different vendors. A department
1142
with the possibility of such a potential failure should identify conditions under which such events
1143
can happen and design in-house preventive strategies (checklists chief among them).
43
1144
Rank
#64
RPN
200
Step#
92
Process
Step
8. Tx
2. Enter prescription + planning constraints
Planning
FM: 1. Incomplete or incorrect set of objectives and constraints
1145
1146
Addressed with rank 47, step 96.
1147
Rank
#65
RPN
200
Step#
138
Process
9. Plan
Approval
FM: 1. Wrong patient
Step
2. Completion of formal prescription after
planning
1148
1149
This FM is very similar to rank 7, step 137 and rank 36, step 139, and QM for this error is the
1150
same.
1151
Rank
#66
RPN
196
Step#
54
Process
Step
7. RTP
Import images into RTP database
Anatomy
FM: 4. Incorrect 3D transformation of image dataset to Tx coord system.
Wrong orientation/scale
1152
1153
This FM is the second of six FMs to appear for this step, and much of the QM for rank 11 step
1154
40 will directly also address this FM. However, this FM is generally a technical failure, as
1155
opposed to a decision and documentation problem as in the Rank 25 FM. QC during the
1156
registration and import process, plus commissioning of the import and registration process before
1157
any new imaging source or protocol is used, can address the technical issues associated with this
1158
error. Review of the registration during planning can be performed by combining images (or
1159
segmented information obtained from the images), and routine use of these types of displays can
1160
help identify scale and orientation problems on a case-by-case basis.
1161
Rank
#67
RPN
195
Step#
90
Process
Step
8. Tx
2. Enter prescription + planning constraints
planning
FM: 1. Incomplete or incorrect set of objectives and constraints
1162
44
1163
This FM relates to inadequately defined treatment planning procedures and guidelines and as
1164
such is best addressed by developing well-defined treatment planning procedures and guidelines.
1165
Otherwise, discussions for rank 47, step 96; rank, 51 step 110; and rank 83, step 89 apply to this
1166
FM as well.
1167
Rank
#68
RPN
194
step#
21
Process
3. CT/Sim
Step
Physically marking isocenter setup point (e.g.
tattoos) on patient and/or immobilization
casts/masks/etc marked as needed
FM: 2. Inadequate process for marking/reproducing isocenter relative to
treatment planning policies and/or verification process so that setup process
sigma is too large for the PTV margin assumed by planning process
1168
1169
Addressed with rank 58, step 19.
1170
Rank
#69
RPN
193
Step#
63
Process
Step
7. RTP
PTV construction
anatomy
FM: 1. PTV not specified or target coverage specified relative to CTV
1171
1172
Though the reasons for this error in PTV construction are different than the earlier target
1173
problems (rank 2, step 58; rank 5, step 59; and rank 6, step 65), the main QA issues are the same.
1174
A check of the PTV and the margin information used for its construction has been added to the
1175
anatomy QA check (Example Checklist 2 in Table V) to mitigate this FM.
1176
1177
Rank
#70
RPN
192
Step#
141
Process
9. Plan
Approval
FM: 4. Wrong plan
Addressed with rank 36, step 173.
Step
2. Completion of formal prescription after
planning
Rank
#71
RPN Step#
Process
Step
191
20
3. CT/Sim
Physically marking iso setup point
FM: 1. Error in physically marking or documenting isocenter setup point (e.g.,
tattoos) on patient and/or immobilization casts/masks/etc.
1178
1179
1180
Addressed with 58, step 19.
45
1181
Rank
#72
RPN
191
Step#
84
Process
8. Tx
planning
FM: 2. Incorrect ROI volumes
Step
1. Specify ROI for optimization process
1182
1183
Volume calculations for regions of interest (ROI) can be quite variable as they are a function of
1184
contour data, methods used, attention to detail, as well as being dependent on the algorithms
1185
used. Commissioning of this feature, especially over the entire range of structure types where
1186
this information will be used clinically, is crucial. During planning, it is also possible to review
1187
volume results, to make sure organ volumes are within the normal range for each organ.
1188
Rank
#73
RPN
190
Step#
38
Process
Step
5. Transfer images Transfer secondary (MRI, PET) datasets
and other
DICOM data
FM: Correct patient selected but incorrect study set selected (wrong site, date,
technique, etc.)
1189
1190
In this FM an incorrect secondary dataset is used to provide additional information for definition
1191
of patient anatomy and planning geometry.
1192
inevitably involves QC measures and independent checks by the physician or medical physicist.
1193
Sometimes, identification of an incorrect MR sequence is possible in routine clinical processes,
1194
provided a relatively high level of experience exists among those using the images. However,
1195
identification of incorrect studies or similar discrepancy may be very difficult or impossible
1196
through a simple review process. As such, this FM requires a QC measure where each imported
1197
secondary dataset has an independent verification, especially for situations where multiple
1198
studies exist. The exact nature of the checks depends on the institutional organization of the PAC
1199
and method of exchanging images.
This is a difficult FM to address and almost
1200
Rank
#74
RPN
189
Step#
180
Process
11. Day 1 Tx and
Day N Tx
FM:3. Incorrect Tx data
1201
46
Step
Gather patient Tx info
1202
This failure can have many causes, including human failure (this FM), software failure (rank
1203
127, step 179) and hardware failure (rank 178, step 178). The Day 1 Treatment check (Example
1204
Checklist 4 in Table V) is the most important factor in avoiding incorrect treatment data, as
1205
everything needed for the daily treatment of the patient should be checked at that procedure.
1206
Attention to the daily treatment process by the therapists, control of interruptions, and applying
1207
techniques to avoid inattention is also necessary. Finally, the periodic physics chart check and
1208
computer record check should verify any changes in the treatment parameters as well as
1209
confirming that the correct parameters were used initially.
1210
Rank
#75
RPN
189
Step#
150
Process
10. Plan
Prep
FM: 2. Wrong prescription
Step
Specification of Tx course
1211
1212
This FM results in the wrong prescription tied to a treatment course.
This error can be
1213
significant, since it can result in the wrong dose being delivered to a patient due to the confusion
1214
between courses and what prescription should be fulfilled. QM to prevent this kind of error
1215
involves a well-defined process for the way the download of information is prepared and
1216
performed.
1217
confirmation that this step is performed correctly.
The physics plan check (Example Checklist 3 in Table VI) should include a
1218
Rank
#76
RPN
188
Step#
171
Process
Step
10. Plan
10. Download complete delivery plan to
Prep
delivery system
FM: 1. Incorrect plan info into delivery system. 2. Connect wrong patient/plan
in RTP with wrong patient/plan in Tx delivery system
1219
1220
This crucial FM is discussed in rank 15 step 168 along with a number of other failure modes.
1221
Any error downloading information into the delivery system can lead to systematic
1222
mistreatments, and must be prevented by careful confirmation of all the information downloaded.
1223
Rank
#77
RPN Step#
Process
187
183
11. Day 1 Tx
FM: 1. Incorrect Tx isocenter
Step
Position patient for Tx
1224
47
1225
Addressed with rank 60 step 181.
1226
Rank
#78
RPN
185
Step#
142
Process
Step
9. Plan
2. Completion of formal prescription after
approval
planning
FM: 5. Premature signaturization
1227
1228
In this failure mode, two approved plans (e.g . large field and boost) are in the delivery system
1229
database, creating the possibility that the therapist might select the wrong plan for daily
1230
treatment or even treat both plans on the same day. A clinical example of similar problems,
1231
along with one department’s solution, was recently described32. It is crucial that the plan
1232
download application or the treatment management system forces the delivery system to make an
1233
association between a plan and the schedule for daily treatment sessions. To prevent the incorrect
1234
scheduling of the sequence of treatments and plans, the pretreatment QA check (Example
1235
Checklist 3 in Table VI) should involve a comparison of the fractionation schedule exported to
1236
the delivery system with the original written prescription. The boost vs. large field ambiguity
1237
can be eliminated by requiring that the treatment management system (TMS) allows and
1238
enforces sequential scheduling of different plans. For delivery systems that do not allow this
1239
decision to be locked down, a prescription and daily treatment record separate from the delivery
1240
system should be used, so that the therapists can QC their plan selection choices on a daily basis.
1241
The wrong patient pathway can be blocked by setting up appropriate delivery system software
1242
interlocks as discussed in the rank 4 step 48 discussions.
1243
Rank
#79
RPN
185
Step#
7
Process
Step
2. Immobilization Patient positioning/immobilization
and
appropriate to Tx
positioning
FM: Faulty execution of immobilization technique.
Incorrect position.
Inadequate fit. Incorrect indexing. Erroneous documentation
1244
1245
Ideally, the physician would be closely involved in the construction of the immobilization device
1246
and the subsequent simulation, but often that ideal is not realized. For this reason, two likely
1247
causes of immobilization failures are 1) insufficient communication from the physician to the
1248
therapist regarding the intended immobilization and patient set-up, and 2) therapist omission to
48
1249
perform necessary features of the immobilization, to carry them through to simulation, or to
1250
record important information.
1251
1252
The best approach to avoid both problems entails use of a department-specific form. The form
1253
requires the physician to fill in blanks to select from among common departmental
1254
immobilization options and special instructions, as well as blanks for the therapist to check off or
1255
enter information when completing the simulation. The likelihood of failures in these two steps
1256
can also be reduced by written protocols for common disease sites detailing the immobilization
1257
materials, treatment aids and body positions and the scanning protocol and anatomical
1258
information to be acquired. It is important that new staff be trained to achieve competency in
1259
these standard protocols and that the protocols be updated as department procedures change.
1260
Departments should encourage physicians to answer therapists’ questions promptly regarding
1261
unusual situations that arise regarding patient setup (e.g., answer pages). Other causes of failure
1262
involve inadequate materials or tools, lack of evaluation of new immobilization materials or
1263
devices (i.e., commissioning of the device), and lack of training for involved personnel in the use
1264
of the devices.
1265
routinely occur during the construction of immobilization devices can prevent many problems in
1266
creation of these devices and their clinical use.
Attention to training, commissioning, and planning for the situations that
1267
Rank
#80
RPN
183
Step#
55
Process
7. RTP
Anatomy
FM: 5. Gray scale conversion
Step
Import images into RTP database
1268
1269
Addressed with rank 11 and checks in Example Checklist 2 in Table V.
1270
Rank
#81
RPN
182
Step#
196
Process
Step
12. Day N
Patient setup for Tx
Tx
FM: Shifts of isocenter established at Day 1 (or at last correction) not applied or
incorrectly applied
1271
1272
Another failure mode for Day N occurs if the isocenter shift determined at Day 1 is not applied.
1273
For modern accelerators, couch coordinates that embody necessary shifts are often “captured”
49
1274
and verified as part of the Day 1 set-up, and changes in couch position outside preset tolerances
1275
require conscious override. This risk renews itself since it is standard practice to change the
1276
isocenter corrections based on periodic (often weekly) portal imaging or more often if IGRT or
1277
adaptive therapy techniques are used. Addressed with rank 58 step 19 and rank 60 step 181.
1278
Rank
#82
RPN
181
Step#
32
Process
Step
5. Transfer images and Transfer primary (CT) dataset
other DICOM data
FM: Incorrect CT dataset associated with patient
1279
1280
This failure mode results from an incorrect CT dataset being associated with the patient’s
1281
treatment plan. The most likely scenario is that a CT dataset for a wrong patient is exported or
1282
imported in to the treatment planning system. While most TPSs provide a comparison of the
1283
patient name in the TPS with the name obtained from the DICOM image header and display a
1284
warning when a mismatch occurs, it is possible through human error, confusion due to similar
1285
patient names, or software or typing errors for an incorrect match to be made. If the user is able
1286
to import the wrong patient’s dataset into the planning system, this error may be very difficult to
1287
detect later if the scan is for the same site and similar patient geometry. In some situations, the
1288
error may be readily discoverable through routine clinical processes by obvious mismatches
1289
between the dataset and other relevant treatment information and work created during treatment
1290
planning. If a patient has multiple CT datasets and if a wrong dataset is imported for treatment
1291
planning, such an error may be impossible to discover through routine clinical processes. The
1292
relatively high rank for this FM indicates that it may be best handled with QC during plan import
1293
and mandatory verification of match between the patient and data. Again, the exact nature of the
1294
checks depends on the institution’s system for handling image sets and the validation routines in
1295
the treatment planning system.
1296
Rank
#83
RPN
181
Step#
89
Process
8. Tx planning
Step
2. Enter prescription + planning
constraints
FM: 1. Incomplete or incorrect set of objectives and constraints
1297
1298
Addressed with rank 45 step 93.
1299
50
Rank
#84
RPN
181
Step#
45
Process
Step
6. Initial Tx Suggested initial guidelines for beam angles,
Planning
energies, machine
Directive
FM: Specified inappropriately or incompletely. Appropriate request not
followed
1300
1301
Addressed with rank 11 step 40.
1302
Rank
#85
RPN
180
Step#
86
Process
Step
8. Tx
2. Enter prescription + planning constraints
planning
FM: 4. ROI expansion outside or close to the outer skin contour
1303
1304
Extension of target volumes into the air can create dose optimization difficulties, suboptimal
1305
plans, and potential hot spots in some planning systems. In such systems, it is important to
1306
ensure that target volumes do not extend beyond the skin or are systematically retracted from the
1307
skin by a certain distance to avoid problems at optimization and plan evaluation. Understanding
1308
of how the planning system handles targets that extend into air should be acquired at
1309
commissioning and should form the basis of the department’s procedures in such situations.
1310
Rank
#86
RPN
180
Step#
195
Process
Step
12. Day N
Patient setup for Tx
Tx
FM: Immobilization aids incorrectly used
1311
1312
Addressed with rank 58 step 19.
1313
Rank
#87
RPN
180
Step#
109
Process
8. Tx
planning
FM: 2. Wrong on/off choice
Step
6. Heterogeneity corrections
1314
1315
This potential human failure is similar to one of the failures in rank 30 step 108, and has been
1316
discussed in detail there. Standardized procedures and a physics plan review check are major
1317
parts of the prevention of this error (see the treatment plan check, Example Checklist 3 in Table
1318
VI). Most modern planning systems are sufficiently sophisticated to document which correction
51
1319
is applied to a particular plan, simplifying discovery of this failure with a simple check at the end
1320
of planning.
1321
Rank
#88
RPN
180
Step#
8
Process
Step
2. Immobilization Radiological properties of positioning aids
and
are known to planners (e.g. attenuation,
Positioning
skin-sparing)
FM: Properties of device not consistent with accurate dose delivery or artifactfree verification imaging. Inadequate procedures
1322
1323
This FM is addressed by commissioning of new devices used for positioning or immobilization,
1324
documentation of their radiological properties and training of simulator personnel and physicians
1325
in device characteristics.
1326
Rank
#89
RPN
179
Step#
99
Process
Step
8. Tx
3. Setup Tx fields (machine, energy, MLC,
planning
beam angles, etc)
FM: 4. Poor selection of beam energy
1327
1328
Suboptimal selection of treatment machine, energy, and other fundamental treatment parameters
1329
can be difficult to detect. Progenitor causes include lack of communication between the
1330
physician and scheduler or choice by a physician or planner of a particular energy when a
1331
different energy could have given a better dose distribution. The failure mode is addressed
1332
through site-specific protocols and by physics plan QC (Example Checklist 3 in Table VI.) This
1333
FM is related to rank 175 step 98.
1334
Rank
#90
RPN
179
Step#
81
Process
Step
8. Tx
1. Specify ROI for optimization process
planning
FM: 1. Incorrect classification of a structure as overlapping and non-overlapping
1335
1336
This FM is closely related to rank 33 step 80, except that in this case the overlap regions between
1337
ROIs are mishandled due to software error. The FM must be addressed first with appropriate
1338
commissioning of this feature, followed by determination of the correct protocols for use in
1339
various clinical situations. Finally, as described in rank 33 step 80, this FM should be prevented
1340
through QC during planning by ensuring that planned dose distributions appear reasonable and
52
1341
are consistent with past experience. QA at the end of the anatomy definition and planning
1342
processes (Example Checklists 2 and 3 in Tables V and VI) can also prevent errors here.
1343
Rank
#91
RPN
179
step#
125
Process
Step
8. Tx
11. Evaluate leaf sequences
Planning
FM: 2. Incorrect evaluation (i.e. ok when not ok, etc)
1344
1345
In this failure mode a deviation between the usual (standard) MLC sequence protocol and the
1346
sequence planned is not detected. The differences can include use of dMLC vs. sMLC or
1347
allowing the leaf abutment to walk through the field where departmental policy is that abutting
1348
leaves that are not shaped to deliver fluence should be parked under the jaws. A completely
1349
incorrect MLC sequence could be allowed due to inattention or failure to review. The potential
1350
errors that can result from wrong fluence distributions, and the difficulty in finding an error once
1351
an IMRT treatment has been approved for clinical use, make this an important issue. This FM is
1352
best addressed through the physics check of the treatment plan and careful commissioning, as
1353
discussed in rank 54 step 120.
1354
Rank
#92
RPN
179
Step#
156
Process
Step
10. Plan
3. Delivery protocols
Prep
FM: Inappropriate protocol (viz. tolerance table, wrong use of automation)
1355
1356
QM for this failure mode has been discussed in rank 15 step 168, as well as other FMs for the
1357
plan preparation step.
1358
Rank
#93
RPN
177
Step#
105
Process
8. Tx
planning
FM: Error in dose calculation
Step
5. Calculate the dose to optimization points
and also the full dose distribution
1359
1360
See rank 30 step 108 for discussion. The only difference between this FM and that at step 108 is
1361
that rank 30 discusses heterogeneity dose corrections and this FM discusses dose calculation in
1362
general. Commissioning, QC during planning, and QA at the end of the planning are all very
1363
analogous.
53
1364
Rank
#94
RPN Step#
Process
Step
177
12
3. CT/Sim
Immobilized patient setup on CT simulator
FM:2. Incorrect/inadequate immob, aids, incorrectly applied resulting in
abnormally large random setup error
1365
1366
There are a number of opportunities in downstream steps to detect failures in immobilization and
1367
simulation before they adversely affect treatment, though correction may cause inconvenience to
1368
patient and staff (e.g. re-simulation and re-planning). A treatment planner usually can quickly
1369
detect steps 14 and 15 (scan technique, volume scanned), missing information (step 22) and tardy
1370
data transfer (step 24). The localization procedure at the initial, and even subsequent treatments,
1371
also provides some protection from some major errors in simulation, a fact reflected in their
1372
detectability rankings.
1373
become apparent through standard-of-care treatment procedures. In this case, the larger random
1374
setup errors might eventually call attention to the problem but possibly only after numerous
1375
fractions. One possible approach to detecting such errors would be imaging more frequently.
1376
Use of standardized protocols and a simulation form helps to reduce the frequency of corrections
1377
that require repeated effort or the more serious consequences of immobilization or simulation
1378
failures. Attention to the adequacy of accessory devices during the physics check at the initial
1379
treatment can also intercept problems.
However, some simulation and immobilization failures would not
1380
Rank
#95
RPN
177
Step#
26
Process
4. Other pre-Tx
imaging for CTV
localization
FM: Special requirements not respected
Step
2. Patient advised of special
requirements (e.g. fast before FDG
PET)
1381
1382
Addressing this FM requires attention to the process used to notify, schedule, and plan imaging
1383
procedures, especially those that may happen outside the control of radiation oncology people
1384
(e.g., MR, PET, etc.). A well-defined standard protocol for all these activities, developed through
1385
collaboration between the relevant persons and departments, can help minimize this error, as can
1386
training for the staff that performs those activities.
1387
Rank
RPN
Step#
Process
Step
54
#96
176
34
5. Transfer imaged Transfer primary (CT) dataset
and other
DICOM data
FM:
Data incompatibility, e.g., Image mirroring (chirality) or patient
orientation mismatch Image interpretation problems (pixel pitch/image size
errors). See TG53 for other mechanisms
1388
1389
This FM is addressed by recommendations of TG5321 and TG6631, including the
1390
recommendations that 1) CT imaging for treatment planning should include orientation markers
1391
embedded directly in the CT couch, and 2) import of patient data into the TPS should include
1392
inspection of patient anatomy for correct laterality identification, orientation, and readily visible
1393
artifacts and deformations in addition to periodic QA measures.
1394
Rank
#97
RPN
175
Step#
121
Process
Step
8. Tx
10. Run leaf sequencer to create deliverable
Planning
plan
FM: 5. Wrong machine configuration
1395
1396
In this FM an undeliverable leaf sequence is created due to poor machine commissioning and
1397
configuration. This FM is addressed thorough system commissioning by an experienced person
1398
who is well trained regarding TPS configuration requirements and characteristics of a particular
1399
delivery system. Extensive creation and delivery of test plans should be performed to validate
1400
deliverability of clinically relevant treatment delivery combinations. See also recommendations
1401
in Rank 54 step 120.
1402
Rank
#98
RPN Step#
Process
Step
174
177
11. Day 1 Tx Gather patient Tx info
FM: 2. Incorrect Tx plan for patient
1403
1404
Addressed with rank 74 step 180.
1405
Rank
#99
RPN
172
Step#
27
Process
Step
4. Other pre-Tx
3. Patient setup for imaging
imaging for CTV
localization
FM: Poor positioning. RT immobilization not used
1406
55
1407
Addressed with rank 95 step 26.
1408
Rank
#100
RPN
171
Step#
126
Process
Step
8. Tx
12. Transfer sequencer results to RTP for final
planning
dose calc (if needed)
FM: RTP (dose calc algorithm) gives wrong intensities
1409
1410
Addressed with rank 12 step 106, rank 54 step 120, and rank 51 step 110.
1411
Rank
#101
RPN
171
Step#
51
Process
Step
7. RTP
Import images into RTP database
Anatomy
FM: 1. Wrong patient’s images selected or imported
1412
1413
This is similar to FMs rank 82 step 32, rank 73 step 38, and rank 76 step 171. Most modern
1414
commercial systems have built-in protection against such failures. However, for older systems
1415
or in-house systems without such protection, this FM is generally a human failure, as most
1416
systems that import images display patient name and registration number or other unique
1417
identifying information as the import process proceeds.
1418
misspelling, inattention, and other similar problems can cause this failure. QC during the
1419
process is thus crucial, because there may be virtually no way to identify that the images are an
1420
incorrect image set if it is not properly identified at this point. Therefore, it is crucial to perform
1421
this step with great care, and for the treatment planner to investigate any unusual or inconsistent
1422
information related to the images, since those subtle errors may hint at some bigger problem that
1423
should be investigated. Once imported, detection of these errors when the patients do not differ
1424
in obvious ways becomes exceedingly difficult. Future improvements in system design should
1425
eliminate this failure mode.
However, similar names, name
1426
Rank
#102
RPN
169
Step#
101
Process
Step
8. Tx
3. Setup Tx fields (Machine, energy, MLC,
planning
gantry, angles, etc)
FM: 6. Incorrect selection of isocenter
1427
1428
In this FM, the treatment plan isocenter is inadvertently different from the marked isocenter for
1429
whatever reason (e.g., instructions not clear, communication not established between treatment
56
1430
planner and therapists) and no couch shift is specified in the setup instructions. Incorrect
1431
treatment isocenter can potentially lead to serious delivery errors if setup fields are created for
1432
the isocenter marked on the patient but the plan is delivered to the incorrect isocenter. IMRT
1433
plan QC must include verification that the setup fields and other images and data used for patient
1434
position verification are consistent with the treatment plan and that they all reference the same
1435
anatomical location. The physician is better able to detect wrong irradiation sites and physician
1436
review of all Day 1 imaging before further (or any treatment) should be mandatory. Otherwise,
1437
there is a potential that patient positioning appears correct but that a very wrong location is
1438
irradiated. Setup imaging and other such information is an appropriate on-going QC test to make
1439
sure that the isocenter used for setup is correct for each treatment session.
1440
Rank
#103
RPN
169
Step#
17
Process
3. CT/sim
Step
Documentation or unusual sim attributes (in
print or software) e.g. unusual patient
positioning
FM: Sim attributes not correctly documented
1441
1442
As in other FMs for the simulation process, carefully following standardized procedures and
1443
protocols for the simulation of given clinical sites will avoid many of these errors. Use of a
1444
Simulation Form can also help prevent failures. Finally, review of the results of the entire
1445
simulation by the appropriate staff members (i.e., physician, medical physicist, dosimetrists or
1446
therapists) at the end of the procedure can supply any missing information.
1447
Rank
#104
RPN
168
Step#
69
Process
Step
7. RTP
Boolean combination of delineated structures
Anatomy
FM: Boolean combination structure representation incorrect due to software
error or failing to observe algorithm limits
1448
1449
This FM is discussed in rank 2 step 58, which recommended using standardized protocols and
1450
following Example Checklist 2 in Table V to prevent this failure.
1451
commissioning of any clinically -used features is essential.
1452
algorithm, as important details of the implementation may only show up during extreme or
1453
unusual situations.
57
In addition, detailed
It is important to stress the
1454
Rank
#105
RPN
168
Step#
88
Process
Step
8. Tx
1. Specify ROI for optimization process
planning
FM: 4. ROI expansion outside or close to outer skin
1455
1456
See rank 85 step 86 for discussion. The plan directive or standard protocol for the clinical site
1457
should determine the method of expansion to be used for each case.
1458
Rank
#106
RPN
167
Step#
6
Process
Step
2. Immobilization Patient positioning and immobilization
and positioning
appropriate to Tx
FM: Faulty execution of immobilization technique.
Incorrect position.
Inadequate fit. or indexing. Erroneous documentation
1459
1460
Addressed with rank 79 step 7.
1461
Rank
#107
RPN
167
Step#
104
Process
Step
8. Tx
4. Setup dose calc parameters
planning
FM: 2. Poor parameters selected
1462
1463
Addressed with rank 51 step 110.
1464
Rank
#108
RPN
165
Step#
91
Process
Step
8. Tx
2. Enter prescription + planning constraints
planning
FM: 1. Incomplete or incorrect set of objectives or constraints
1465
1466
Addressed with 47 step 96 and rank 51 step 110. Verification that all of the necessary structures
1467
are correctly contoured and included correctly in the optimization should be included in the plan
1468
QC process, as in the Example Checklist 3 in Table VI treatment planning check. .
1469
Rank
#109
RPN
164
Step#
136
Process
9. Plan
approval
FM: 2. Wrong plan
Step
1. Plan ok to go to Tx
1470
58
1471
This is a hardware/software error version of the rank 20 step 135 in which incorrect, incomplete,
1472
or corrupted plan data are stored on the delivery system. The physics plan check (Example
1473
Checklist 3 in Table VI) and patient-specific dose verification (IMRT QA) should intercept this
1474
error.
1475
Rank
#110
RPN
164
Step#
133
Process
9. Plan
approval
FM:1. Wrong patient
Step
1. Plan ok to go to Tx
1476
1477
In this error pathway, a dosimetrist attempts to export the wrong patient’s plan to the delivery
1478
system. See the rank 7, step 137 discussion. Software interlocks on commercial and modern in-
1479
house systems should intercept these errors.
1480
Rank
#111
RPN
163
Step#
144
Process
10. Plan
prep
FM: 1. Bad info entered
Step
1. Entry of demographic info
1481
1482
Addressed with 15 step 168.
1483
Rank
#112
RPN
163
Step#
176
Process
Step
11. Day 1 Tx and Gather patient Tx info
Day N Tx
FM: 1. Incorrect patient records used (wrong chart or database file selected)
1484
1485
Addressed with rank 74 step 180.
1486
Rank
#113
RPN
163
Step#
210
Process
Step
11. Day 1 Tx and Tx delivered
Day N Tx
FM: Gantry or other linac hardware (e.g. OBI) collides with patient
1487
1488
Collision between the gantry (or attached hardware) and the patient is a very real hazard.
1489
Treatments including unusual couch angles or other non-standard positions of the equipment or
1490
patient present an increased risk. Counting on observing the clear movement of the treatment
59
1491
unit on a video monitor poses risks because the angle of the image may not permit appreciation
1492
of imminent collision and due to a lapse in the attention of the operator. Departments should
1493
have a policy requiring verification of clearance with the operator in the room rotating the unit
1494
through the range of treatment beams before moving the gantry from the treatment console.
1495
Because limiting an activity with policies is not a strong barrier to the activity, it would be more
1496
effective to also have collision detection systems. These are available for some machines with
1497
on-board imaging capabilities and often were part of radiographic simulators, but have not
1498
become a standard part of most accelerators. Facilities purchasing accelerators should include
1499
collision detection in the purchase specification, even if not currently available, so the vendors
1500
recognize the need to provide this feature, and make it available in the future. Until such time as
1501
collision avoidance becomes a normal part of accelerators, establishing a policy for verification
1502
of clearance before leaving a treatment room if gantry movement will be performed from the
1503
treatment console, and watching for patient movement, will have to serve for the preventive
1504
measure at this time.
1505
Rank
#114
RPN
161
Step#
170
Process
Step
10. Plan
10. Download complete delivery plan to
Prep
delivery system
FM: 1. Incorrect plan info into delivery system: 2. Connect wrong patient/plan in
RTP with wrong patient/plan in Tx Delivery system
1506
1507
Addressed with rank 76 step 171.
1508
Rank
#115
RPN
161
Step#
117
1509
Process
Step
8. Tx
10. Run leaf sequencing to create deliverable
Planning
plan
FM: 1. Error in sequencer
In this FM, the TPS creates an undeliverable leaf sequence due to an error in the sequencer. QM
1510
to prevent errors in sequencing has been discussed in rank 54 step 120 above. Commissioning is
1511
an especially important activity if possible errors in the sequencer are going to be found.
1512
Commissioning should ensure that the sequencer system functions correctly over the whole
1513
range of clinically useful configurations.
1514
Rank
RPN
Step#
Process
Step
60
#116
160
11
3. CT/sim
Immobilized patient setup on CT sim
FM: 1. Incorrect/inadequate immobilization; immobilization aids incorrectly
applied, resulting in large >3sigma setup error
1515
1516
Addressed with rank 94 step 12.
1517
Rank
#117
RPN
158
Step#
167
Process
Step
10. Plan
9. Prepare e-chart
Prep
FM: 1. Incorrect Tx info. 2. Wrong Rx. 3. Wrong patient/plan
1518
1519
Addressed with rank 15 step 168.
1520
Rank
#118
RPN
158
Step#
35
Process
5. Transfer images
and other
DICOM data
FM: 1. Incorrect patient
Step
Transfer secondary (MRI, PET) datasets
1521
1522
See rank 82 step 32 for discussion.
1523
Rank
#119
RPN
158
Step#
146
Process
Step
10. Plan
1. Entry of demographic info
Prep
FM: 2. Failure to link to external database
1524
1525
Addressed with rank 15 step 168.
1526
Rank
#120
RPN
157
FMEA#
124
Process
Step
8. Tx
11. Evaluate leaf sequences
Planning
FM:1. Unintentional modification of sequences
1527
1528
Unintentional modification of the MLC leaf sequence can be a major problem because it can
1529
result in delivery errors that can either over-treat or under-treat some patients. This type of error
1530
would most generally be tied to software, hardware, network, or other system-related failures,
1531
though human errors were also involved. QM for this failure has been discussed in rank 91 step
61
1532
125. The staff running the treatment must be constantly watchful and attentive, to be aware of
1533
unusual or unexpected things happening. Additional testing of MLC sequencing can include
1534
fluence-pattern review as part of the pre-treatment physics check.
1535
Rank
#121
RPN
157
Step#
103
Process
Step
8. Tx
14. Setup dose calc parameters
Planning
FM: 1. Wrong parameters selected
1536
1537
In this FM, for example, an incomplete dose calculation is performed by setting the size of the
1538
dose calculation matrix smaller than the target volume or volume of organs at risk. Incomplete
1539
dose calculation then leads to plan evaluation errors (often due to incomplete DVH results). To
1540
avoid this FM, IMRT plan evaluation should always include isodose review with special
1541
concentration on lower isodose levels. See rank 10 step 127 for more detail.
1542
Rank
#122
RPN
157
Step#
103
Process
8. Tx
Planning
FM: 1. Error in sequencer
Step
10. Run leaf sequencing to create deliverable
plan
1543
1544
Addressed with rank 54 step 120 and rank 115 step 117.
1545
Rank
#123
RPN
156
Step#
71
Process
7. RTP
Anatomy
FM: Density map incorrect
Step
Converting CT image intensity to density
1546
1547
This is the first FM directly relating to the use of CT density information (though density effects
1548
on heterogeneity-corrected dose calculation algorithms were discussed in rank 30 step 108).
1549
Commissioning is a crucial necessity for understanding and confirming the correct use of CT
1550
density. However, appropriate review of some test clinical cases is also an important component
1551
of QM for this failure mode, given the large variations of anatomical shapes and densities that
1552
are encountered in the clinic. Individual CT scan sets can show great variations from patient to
1553
patient, which can stress any commissioning and testing regime. Periodic evaluation of the
1554
conversion for each CT unit involved in treatment planning is also recommended, although the
62
1555
periodicity of such testing has yet to be determined. In the absence of any data of the likelihood
1556
of changes in the conversion between CT-units and density, evaluation with the annual imaging
1557
QA on the CT unit seems appropriate. Special consideration should be given to the quantification
1558
of high-Z implanted material such as tungsten.
1559
Rank
#124
RPN
155
Step#
4
Process
Step
2. Immobilization Patient positioning/immobilization
and Positioning
appropriate to Tx
FM: Inappropriate immobilization method. Incorrect position. Wrong choice of
materials or accessories.
1560
1561
Addressed with rank 79 step 7.
1562
Rank
#125
RPN
154
Step#
200
Process
Step
12. Day N
Place blocks
Tx
FM: Blocks used in addition to MLC are not placed.
1563
1564
Devices needed for daily treatment that cannot be directly tied into the computer control system
1565
of the delivery system require extra attention during the treatment process if they are to be
1566
remembered and used correctly. The best way to avoid errors is to use a formal procedure for
1567
the treatment process that forces the documentation of the use of each of these devices, thereby
1568
acting as a reminder as well as documentation of the use of the device. See the discussion at
1569
rank 8, step 205 and the text of Sec. 9 E of the main body of the TG100 protocol, rank 24.
1570
Rank
#126
RPN
154
Step#
56
Process
7. RTP
Anatomy
FM: 6. Distortion correction
Step
Import images into RTP database
1571
1572
This particular FM deals with inaccurate correction of distortions, until recently mainly an issue
1573
for MR scans. However, current research efforts are starting to use so-called deformable image
1574
registration and other similar technology to handle “distortions” or artifacts created by patient
1575
motion or change over time. There are now many more reasons that incorrect distortions may
1576
become part of the planning information. QM to handle these issues has not been well defined,
63
1577
as these techniques have barely become part of the clinical arsenal of tools. More effort in this
1578
area is certainly needed.
1579
Rank
#127
RPN Step#
Process
153
179
11. Day 1 Tx
FM: 3. Incorrect Tx data
Step
Gather patient Tx info
1580
1581
Addressed with rank 74 step 180.
1582
Rank
#128
RPN Step#
Process
Step
152
188
11. Day 1 Tx Set Tx parameters
FM: 1. Wrong parameters uploaded to treatment unit
1583
1584
Rank 24 step 189 (see Sec. 9 D3 in the main body of the TG 100 protocol), rank 60 step 181,
1585
rank 74 step 180 all have QM relevant to this FM. A check of the parameters for treatment must
1586
be performed as the plan is downloaded, and also at Day 1.
1587
Rank
#129
RPN
152
Step#
87
Process
Step
8. Tx
1. Specify ROI for optimization process
Planning
FM: 4. ROI expansion outside or close to the outer skin contour
1588
1589
Addressed with rank 85 step 86.
1590
Rank
#130
RPN
152
Step#
131
Process
8. Tx
Planning
Step
9. Transfer optimized fluence to leaf sequencer
– if separate system. 1. Correct patient or
plan
FM: Incorrect editing
1591
1592
In this FM, an incorrect plan is exported to a secondary system that creates leaf sequences or
1593
handles transfer of treatment plan leaf sequences to the treatment machine. As with the earlier
1594
FMs related to the leaf sequencing, this FM can be addressed through checks of the leaf pattern
1595
or an irradiation of a phantom with the patient’s plan or by sending a log of delivered leaf
1596
sequences from the treatment machine to an independent dose calculator and comparing results
1597
with the output of treatment planning. This option could be significantly automated and need not
64
1598
require significant clinical resources.
But currently, such systems are mainly homegrown;
1599
vendors are urged to develop commercial solutions.
1600
Rank
#131
RPN
151
Step#
70
Process
7. RTP
Anatomy
FM: Incorrect editing
Step
Editing density map for imaging artifacts e.g.
contrast and bolus
1601
1602
Addressed with rank 123 step 71.
1603
Rank
#132
RPN
151
Step#
134
Process
9. Plan
Approval
FM: 1. Wrong patient
Step
1. Plan OK to go to Tx
1604
1605
Addressed with rank 7 step 137.
1606
Rank
#133
RPN
149
Step#
119
Process
Step
8. Tx
10. Run leaf sequencing to create deliverable
Planning
plan
FM: 3. Sequence file or output incomplete or incorrect
1607
1608
Addressed with rank 120 step 124 and rank 130 step 131.
1609
Rank
#134
1610
RPN Step#
Process
Step
148
185
11. Day 1 Tx Position patient for Tx
FM:3. Incorrect patient orientation
Addressed with rank 3 step 24, rank 60 step 181 and rank 74 step 180.
1611
Rank
#135
RPN Step#
Process
Step
147
175
11. Day 1 Tx Take patient into Tx room
FM: 2, Patient medical condition changes between prescription and treatment
(dental situation, wt changes)
1612
1613
Addressed with rank 3 step 24, rank 60 step 181 and rank 74 step 180.
1614
Rank
RPN
Step#
Process
Step
65
#136
147
49
7. RTP
Create case (define patient in RTP database)
Anatomy
FM: 1. Misidentification (wrong name)
1615
1616
Addressed with rank 15 step 168, as well as rank 4 step 48 and others.
1617
Rank
#137
RPN
146
Step#
82
Process
Step
78. Tx
1. Specify ROI for optimization process
Planning
FM: 1.Incorrect classification of a structure as overlapping or non-overlapping
1618
1619
This FM has numerous causes, but this entry is directly related to software limitations, for
1620
example overrunning the number of overlapping structures that are allowed. Commissioning is
1621
thus crucial to demonstrating and identifying any software weaknesses or limitations. Training
1622
and procedures can then use those identified issues to make sure appropriate decisions are taken.
1623
Further QM for this and related FMs has been discussed in rank 33 step 80.
1624
Rank
#138
RPN
146
Step#
37
Process
Step
5. Transfer images Transfer secondary (MRI, PET) datasets
and other
DICOM data
FM: Data incompatibility, e.g. Image mirroring (chirality) or patient orientation
mismatch Image interpretation problems (pixel pitch/image size errors). See
TG53 for other mechanisms
1625
1626
Transfer and input of secondary or unusual imaging datasets is very important, and should be the
1627
subject of significant commissioning testing for any new planning system, format, imaging
1628
device, any new version of DICOM output, transfer, or input software, as details of the
1629
information contained in the files can easily change or be misinterpreted by transfer or input
1630
software. Human decisions and entries into the scanning system (e.g., labeling the patient
1631
position head-first or feet-first) can also be incorrect, leading to other image interpretation
1632
problems. In addition, non-standard datasets, perhaps obtained from outside the institution, can
1633
also be requested and used for planning. This can be a potentially serious problem that cannot be
1634
addressed during commissioning, as the secondary dataset(s) may come from scanners that have
1635
not gone through a formal RT commissioning process and may lack embedded orientation
1636
markers.
66
1637
1638
Related FMs have been discussed in rank 41 step39, rank 73 step 38, rank 82 step 32, and
1639
particularly rank 96 step 34. Whenever possible, secondary datasets, especially for symmetric
1640
body sites (head, pelvis, extremities) should include orientation markers for verification of
1641
patient laterality. Additionally, secondary datasets should be compared with the planning CT
1642
dataset for correct orientation, scaling, and any other deformations.
1643
imaging datasets are used, the QC process should involve checking the consistency of the new
1644
imaging data with the current anatomical information, and discrepancies or inconsistencies
1645
should be noted or addressed. Without this on-going QC, there is a chance that chirality errors or
1646
image deformations may go undetected, resulting in irradiation of wrong anatomy or site.
Whenever secondary
1647
Rank
#139
RPN
140
Step#
5
Process
Step
2. Immobilization Patient positioning/immobilization
and Positioning
appropriate to Tx
FM: Suboptimal immobilization method. Incorrect position. Wrong choice of
materials or accessories. Poorly fitting mask, Best/Typical Case
1648
1649
Addressed with rank 79 step 7.
1650
Rank
#140
RPN
140
Step#
22
Process
3. CT/Sim
1651
Step
Other setup data acquired and documented in
chart (e.g. caliper or ruler measurements of
“AP setup depth”, setup photos)
FM: 1. Measurements not made. 2. Measurements are incorrect. 3.
Measurements are not clearly documented, e.g., (photos unclear or don’t show
key features of setup such as arm position or setup mark locations)
In some clinical situations, a simple measurement can determine many details about the
1652
treatment plan, and if that measurement is incorrect, the entire plan may be flawed. This
1653
measurement is typically a part of the simulation process, and must be handled just like the
1654
typical information obtained from CT scans.
1655
procedures, training of the staff, planner/physicist awareness of relevant simulation information
1656
and a second check (QC) made during the measurement process.
QM should involve following standardized
1657
Rank
#141
RPN
140
Step#
163
Process
10. Plan
Prep
Step
6. Define, annotate anatomical info to be used
for localization
67
FM: Incorrect anatomy visualization or identification
1658
1659
Incorrectly identifying anatomy used in DRRs or patient setup can lead to incorrect setup at one
1660
or more treatment fractions.
1661
during the task, a QA check of the plan as it is prepared for download to the delivery system
1662
(Example Checklist 3 of Table VI), and a final check of consistency during the Day 1 treatment
1663
process can all help to minimize this error.
Careful documentation upstream in the process, attentiveness
1664
Rank
#142
RPN Step#
Process
140
182
11. Day 1 Tx
FM: 1. Incorrect Tx isocenter
Step
Position patient for Tx
1665
1666
Addressed with rank 60 step 181.
1667
Rank
#143
RPN
139
Step#
190
Process
Step
12. Day N
Patient enters room
Tx
FM: Wrong patient (records upon which Tx will be based correspond to a
different patient)
1668
1669
A mismatch between the patient in the treatment room and the treatment plan to be used is a
1670
major error. As with surgery and other procedures, multiple types of identification checks of
1671
patient and the patient-related information and plan in the treatment unit computer should be
1672
confirmed on a daily basis as part of a time out procedure.
1673
Rank
#144
RPN
138
Step#
50
Process
Step
7. RTP
Create case (define patient in RTP database)
Anatomy
FM: 2. Wrong or non-unique ID
1674
1675
As with many errors in patient or plan misidentification, these errors can, if not found, can lead
1676
to much confusion and mistreatment. QM has been discussed in rank 15 step168, as well as rank
1677
4 step 48, to name some of the higher ranked related FMs. As with many failures of this type,
1678
most modern systems have checks to prevent such actions.
1679
68
Rank
#145
RPN
137
Step#
61
Process
Step
7. RTP
Delineate GTV/CTV
Anatomy
FM: 4. Topologically inconsistent contours, overlaps, loops, etc.
1680
1681
Addressed with rank 2 step 58 as well as other anatomy-related errors, such as rank 29 steps 66,
1682
rank 33 step 80, rank 46 step 68, rank 59 step 205 and rank 90 step 81. This particular FM is a
1683
technical problem, but could have the same consequences as the other errors.
1684
Rank
#146
RPN
136
Step#
75
Process
Step
7. RTP
Creation of 3-D anatomical representations
Anatomy
from contours
FM: 3-D voxel representation problems
1685
1686
As with many anatomical representations, both commissioning and routine QC during planning
1687
for each case must be used to prevent these types of failures. In particular, review of the final
1688
anatomical model (in 3-D, to verify all representations) is an important step, as well as
1689
performing a final anatomy check (Example Checklist 2 of Table V).
1690
Rank
#147
RPN
136
Step#
18
Process
3. CT/Sim
Step
Patient position properly represented by
image-transfer software
FM: Unusual patient position not handled by image transfer software (e.g., L
and R labels)
1691
1692
This FM can lead to incorrect patient positioning, wrong side treated, and similar problems. All
1693
image transfer protocols must be commissioned for each patient position label that will be used
1694
clinically. Planner or physicist’s QC of each patient’s positioning information as transferred from
1695
the CT simulation data to treatment planning is a crucial second check.
1696
Rank
#148
RPN Step#
Process
Step
136
174
11. Day 1 Tx Take patient into Tx room
FM: 1. Incorrect patient in the room
1697
1698
Addressed with rank 143 step 190.
1699
69
Rank
#149
RPN
135
Step#
152
Process
Step
10. Plan
2. Specification of Tx course
Prep
FM: 3. Wrong course scheduling etc.
1700
1701
Addressed with rank 75 step 150.
1702
Rank
#150
RPN
135
Step#
154
Process
Step
10. Plan
2. Specification of Tx course
Prep
FM: 4. Fields not ordered correctly (initial, boost)
1703
1704
Addressed with rank 75 step 150.
1705
Rank
#151
RPN
132
Step#
183
Process
Step
12. Day N
Delivery recorded in paper chart (if
Tx
applicable)
FM: Treatment delivery is not recorded in chart
1706
1707
Specific issues for recording vary depending on the specific systems used (computer-controlled
1708
treatment delivery system, paper charts, etc.). When paper charts are used, attention by the
1709
treatment therapists is essential to ensure that the paper record is consistent with what actually
1710
happened. If only manual records are available it can be very difficult to resolve an inconsistency
1711
between therapist memory and the records. The presence of electronic records and at least
1712
weekly physics and therapist chart reviews are important QA. Any event should be investigated
1713
immediately, so that a decision about the corrective action can be made promptly.
1714
Rank
#152
RPN
131
Step#
166
Process
Step
10. Plan
8. Prepare paper chart
Prep
FM: 1. Incorrect Tx info. 2. Wrong Rx. 3. Wrong patient/plan
1715
1716
Addressed with rank 15 step 168 and rank 151 step 183. The paper chart is sensitive to many
1717
typographical and transcription errors so can have many kinds of inaccuracies. The plan check
1718
(Example Checklist 3 of Table VI), physician plan approval step, and Day 1 Tx check are all part
70
1719
of the QA associated with preventing these errors, as is the weekly chart check (Example
1720
Checklist 5 of Table VIII). All of these checks must be employed if paper charts are in use.
1721
Rank
#153
RPN
130
Step#
203
Process
Step
12. Day N
Tx machine and peripheral hardware setup
Tx
for Tx
FM: MLC files (leaf motion) corrupted
1722
1723
This FM has recently been shown to have caused major errors in clinical IMRT treatments, due
1724
to the high severity of these consequences of this problem, if not identified. See discussion in
1725
main article, section 3, rank 153. In principle, this FM can occur on any day of treatment and
1726
would therefore not be detected by typical pre-treatment QA. See also discussion in rank 32 step
1727
207.
1728
Rank
#154
RPN
128
Step#
123
Process
Step
8. Tx
10. Run leaf sequencing to create deliverable
Planning
plan
FM: 7. Error splitting large fields for Varian MLC
1729
1730
Addressed with rank 120 step 124 and rank 130 step 131.
1731
Rank
#155
RPN
127
Step#
29
Process
Step
4. Other pre-Tx
4. Scan performed
imaging for CTV
localization
FM: b. Correct area imaged with wrong protocol
1732
1733
The major problem in this failure mode is that of inconvenience, since inadequate imaging may
1734
force the procedure to be re-done. However, if the protocol is poorly enough handled, contours
1735
drawn based on a poor image might be wrong and lead to poor treatment. Careful attention to
1736
defined protocols for individual clinical sites and sensitivity to the possibility of this type of
1737
problem can likely prevent any significant problem from affecting the patient’s treatment.
1738
Rank
#156
RPN
127
Step#
74
Process
7. RTP
Anatomy
Step
Creation of 3-D anatomical representations
from contours
71
FM: Flawed or erroneous 3-D surface mesh
1739
1740
Addressed with rank 2 step 58.
1741
Rank
#157
RPN
127
Step#
148
Process
10. Plan
Prep
FM: 1. Wrong patient or plan
Step
2. Specification of Tx course
1742
1743
Addressed with rank 75 step 150, rank 15 step 168 and rank 76 step 171.
1744
Rank
#158
RPN Step#
Process
126
192
11. Day 1 Tx
FM: 1. Tx not recorded
Step
Record Tx
1745
1746
Addressed with rank 151 step 183 for paper charts. For electronic charts this failure can happen
1747
either because the computer system was inoperable during the treatment or because of a software
1748
failure. The omission of a treatment would be noticed during either the physics or the therapist
1749
weekly chart check, leading to an investigation of whether the treatment occurred. The RPN for
1750
this failure assumes that such checks are not performed and highlights their importance.
1751
Rank
#159
1752
RPN
125
Step#
140
Process
9. Plan
Approval
Step
2. Completion of formal prescription after
planning
FM: 3. Wrong site
Addressed with rank 36 step 139 and others.
1753
Rank
#160
RPN Step#
Process
124
193
11. Day 1 Tx
FM: 2. Recorded incorrectly
Step
Record Tx
1754
1755
Addressed with rank 151 step 183 and rank 158 step 192.
1756
Rank
#161
RPN
123
Step#
76
Process
Step
7. RTP
Dose point definition (for calc of dose on
Anatomy
random mesh of points for DVH or opt use)
FM: 3-D points distribution errors
72
1757
1758
As with most other anatomy FMs, checks of all anatomical representations at the anatomy check
1759
(Example Checklist 2 of Table V) can minimize these errors.
1760
Rank
#162
RPN
119
Step#
72
Process
Step
7. RTP
Editing anatomical model to conform to
Anatomy
limitations of the TPS
FM: 1. Incorrectly reaching system limits for contours, structures, etc.
1761
1762
As with most other anatomy FMs, careful checks of all anatomical representations at the
1763
anatomy check (Example Checklist 2 of Table V) can minimize these errors.
1764
Rank
#163
RPN
118
Step#
25
Process
Step
4. Other pre-Tx
1. Time of study is compatible with
imaging for CTV
intended use
localizaiton
FM: Study scheduled at bad time relative to simulation or intended use
1765
1766
This FM can be addressed with protocols for scheduling needed tests.
1767
Rank
#164
RPN
117
Step#
201
Process
Step
12. Day N
Tx machine and peripheral hardware setup
Tx
for Tx
FM: File related to Tx corrupted or incorrectly loaded
1768
1769
Addressed with rank 153 step 203 and related testing.
1770
Rank
#165
RPN
112
Step#
13
Process
3. CT/Sim
Step
Patient prepped for imaging session (contrast,
etc)
FM: Contrast not used when it is needed (chronologically follows setup).
Bladder or bowel prep not done (chronologically done before pt comes in)
Breathing control (gating or breath hold) intended but not available at
simulation.
1771
1772
See discussion in rank 35 step 199, which would apply to this failure mode but at the time of the
1773
planning CT.
73
1774
Rank
#166
RPN Step#
Process
Step
110
14
3. CT/Sim
Volume needed for Tx planning scanned
FM: Inadequate quality/Incomplete Images. 1. Wrong scan protocol used (e.g.
wrong slice thickness/separation)
1775
1776
Addressed with rank 155 step 29. This problem can often cause poor plans or inconvenience due
1777
to need to re-do scans.
1778
Rank
#167
RPN
108
Step#
130
Process
Step
8. Tx
16. Evaluate delivery system limitations
Planning
FM: 1. Tx volume inappropriate for delivery system
1779
1780
Patient selection and placement on simulation protocols specific to a treatment site should
1781
prevent this kind of early failure.
1782
Rank
#168
RPN
108
Step#
77
Process
Step
7. RTP
Support of multiple anatomical models (ie,
Anatomy
different body parts) for the same patient
FM: Scans for different patient representations combined or mixed up
1783
1784
Poor bookkeeping or attention to departmental policies on the creation of anatomy and plans for
1785
different regions of the patient can lead to confused and incorrect anatomy and planning
1786
information. Early attention to detail at the simulation and the creation of the patient record can
1787
prevent these kinds of mix-ups.
1788
Rank
#169
RPN
107
Step#
1
Process
Step
1. Patient
Entry of patient data in electronic database or
database
written chart
info
FM: 1. Incorrect patient ID data
1789
1790
Addressed with rank 22, step 3.
1791
Rank
#170
RPN
107
Step#
9
Process
2. Immobilization
74
Step
Patient-specific hardware labeling
and Positioning
FM: Hardware used on wrong patient
(name)
1792
1793
Addressed with rank 79 step 7.
1794
Rank
#171
RPN
106
Step#
111
Process
Step
8. Tx
8. Run optimization
Planning
FM: 1. Tx plan has significant failure meeting planning goals and objectives
1795
1796
Treatment plans can be poor and poor plans need to be recognized and corrected or improved if
1797
possible. Identification of such plans would be during the physics check after treatment planning.
1798
See rank 47 step 96 for relevant discussion.
1799
Rank
#172
RPN
106
Step#
132
Process
Step
8. Tx
16. Evaluate delivery system limitations
Planning
FM: 3. Patient and delivery system collision
1800
1801
Few treatment-planning systems identify potential collision situations, and identifying potential
1802
problems during reviews of the plans becomes very difficult. Interception of these problems
1803
comes at the time of the first treatment run of the fields, assuring clearances. See rank 113 step
1804
210 for discussion.
1805
Rank
#173
RPN
106
Step#
213
Process
Step
12. Day N
Delivery record by delivery system (if
Tx
applicable)
FM: System fails to record Tx due to software bug
1806
1807
Addressed with rank 158 step 126 and rank 151 step 183.
1808
Rank
#174
RPN
105
Step#
98
Process
Step
8. Tx
3. Setup of Tx fields (machine, energy, MLC,
planning
beam angles, etc)
FM: 2. Incorrect or poor selection of Tx machine
1809
1810
Addressed with rank 89 step 99.
75
1811
Rank
#175
RPN
103
Step#
100
Process
Step
8. Tx
3. Setup of Tx fields (machine, energy, MLC,
planning
beam angles, etc)
FM: 5. Incorrect selection of beam energy
1812
1813
Addressed with rank 89 step 99.
1814
Rank
#176
RPN
103
Step#
145
Process
Step
10. Plan
1. Entry of demographic info
Prep
FM: 2. Failure to link to external database
1815
1816
This FM relates to many others with respect to gathering patient data. See rank the discussion at
1817
rank 22 step 3.
1818
Rank
#177
RPN Step#
Process
103
178
11. Day 1 Tx
FM: 3. Incorrect Tx data
Step
Gather patient Tx info
1819
1820
Addressed with rank 74 step 180.
1821
Rank
#178
RPN
102
Step#
115
Process
Step
8. Tx
9. Transfer optimized fluence to leaf sequencer
Planning
(if separate system)
FM: 3. Input data or file corrupted at sequencer
1822
1823
Addressed with rank 54 step 120.
1824
Rank
#179
RPN
102
Step#
158
Process
Step
10. Plan
I4. Prepare DRRs and other localization
Prep
imaging data
FM: Incorrect images (wrong angle, divergence, etc)
1825
1826
Addressed with rank 50 step 159.
1827
Rank
RPN
Step#
Process
Step
76
#180
99
162
10. Plan
6. Define, annote anatomical info to be used in
Prep
localization process
FM: Incorrect anatomy visualization or identification
1828
1829
Addressed with rank 141 step 163.
1830
Rank
#181
RPN
98
Step#
2
Process
Step
1. Patient
Entry of patient data in electronic database or
database
written chart
info
FM: 1. Incorrect patient ID data
1831
1832
Addressed with rank 22, step 3.
1833
Rank
#182
RPN Step#
Process
Step
98
15
3. CT/Sim
Volume needed for Tx planning scanned
FM: Inadequate quality/Incomplete Images. 2. Inadequate volume scanned:
(e.g. doesn’t include all ROIs needed for plan). 3. Wrong location scanned. 5.
Excessive patient motion during scan or other (e.g., metal) artifacts
1834
1835
Addressed such as with rank 155 step 29.
1836
Rank
#183
RPN
96
Step#
151
Process
10. Plan
Prep
FM: 2. Wrong prescription
Step
2. Specification of Tx course
1837
1838
Addressed with rank 75 step 150.
1839
Rank
#184
RPN
93
Step#
153
Process
Step
10. Plan
2. Specification of Tx course
Prep
FM: 3. Wrong course scheduling, etc.
1840
1841
Addressed with rank 75 step 150.
1842
Rank
#185
RPN
91
Step#
157
Process
10. Plan
Step
3. Delivery protocols
77
Prep
FM: Inappropriate protocol (viz., tolerance table, wrong use of automation)
1843
1844
Addressed with rank 92 step 156.
1845
Rank
#186
RPN Step#
Process
Step
89
161
11. Day 1 Tx Tx delivered
FM: Only partial Tx delivered (e.g. 3 of 5 beams)
1846
1847
Partial treatment of a treatment session happens due to machine problems or other issues. The
1848
main failure associated with that event is an incorrect recording or decision about how to deal
1849
with the missing treatment. Each treatment protocol should have a well-defined procedure to
1850
deal with incomplete treatments. The case should be evaluated and changes prepared for the
1851
next treatment session, so physician, therapist, dosimetrist and physicist all know the plan for
1852
handling this situation. The details of the plan depend on the treatment units available, the nature
1853
of the patient’s disease and the likely duration for the machine failure.
1854
Rank
#187
RPN
86
Step#
30
Process
Step
4. Other pre-Tx
5. Scan made accessible for RT
imaging for CTV
planning in timely fashion
localization
FM: Delay in making scan accessible. Images accidentally deleted
1855
1856
This mainly causes inconvenience.
1857
Rank
#188
RPN
86
Step#
155
Process
Step
10. Plan
2. Specification of Tx course
Prep
FM: Fields not correctly ordered (initial/boost)
1858
1859
Addressed with rank 75 step 150.
1860
Rank
#189
RPN
85
Step#
185
Process
Step
12. Day N
Delivery recorded in paper chart (if
Tx
applicable)
FM:Partial delivery incorrectly recorded
1861
78
1862
Addressed with rank 186 step 161, 151 step 183 and rank 158 step 126.
1863
Rank
#190
RPN
83
Step#
112
Process
Step
8. Tx
8. Run optimization
Planning
FM: 3. Optimization fails (does not converge or gives clearly non-optimal
answer)
1864
1865
Addressed with rank 47 step 96.
1866
Rank
#191
RPN
81
Step#
102
1867
Process
Step
8. Tx
Setup of Tx fields (machine, energy, MLC,
Planning
beam angles, beamlet size, etc)
FM: Bad selection of isocenter makes treatment planning hard.
In this failure mode the planner makes a poor selection of isocenter for the plan. As a result it
1868
may be more difficult to achieve the desired dose distribution without delivery of higher than
1869
necessary dose to limiting structures. The physics check of the plan should detect a suboptimal
1870
placement of the isocenter.
1871
Rank
#192
RPN
80
Step#
215
Process
Step
12. Day N
Delivery recorded in paper chart (if
Tx
applicable)
FM: Tx delivery recorded twice
1872
1873
Addressed with rank 186 step 161, 151 step 183 and rank 158 step 126.
1874
Rank
#193
RPN Step#
Process
79
186
11. Day 1 Tx
FM: 1. Images not taken
Step
Localization (portal images, etc)
1875
1876
This failure is easy to detect if part of a well-defined process. A standard protocol and checklist
1877
for new starts (e.g. Example Checklist 4 of Table VII) should address this FM.
1878
Rank
#194
RPN
77
Step#
97
Process
Step
8. Tx
3. Setup treatment fields (machine, energy,
Planning
MLC, beam angle, beamlet size, etc.)
FM: 1. Incorrect selection of Tx technique (e.g., linac vs Tomo)
79
1879
1880
The initial plan directive and physician and physics plan check should both prevent this error
1881
from propagating to treatment, though they will not prevent extra work.
1882
Rank
#195
RPN
75
Step#
149
Process
10. Plan
Prep
FM: 1. Wrong patient or plan
Step
2. Specification of Tx course
1883
1884
Addressed with rank 75 step 150.
1885
Rank
#196
RPN
73
Step#
79
Process
Step
7. RTP
Saving the patient anatomical model
Anatomy
FM: 1. Saving incomplete model or file I/O errors
1886
1887
This error is usually easily detected. However, the check at the end of anatomy definition
1888
(Example Checklist 2 of Table V) should check this.
1889
Rank
#197
RPN
66
Step#
91
Process
Step
8. Tx
10. Run leaf sequencing to create deliverable
planning
plan
FM: 6. Inter-digitization limits incorrect (Siemens, Elekta)
1890
1891
Addressed with rank 97 step 121.
1892
Rank
#198
RPN
65
Step#
165
Process
Step
10. Plan
7. Decide delivery ordering of fields
Prep
FM: Unexpected changes in sequence of fields
1893
1894
This could be detected during a weekly chart check. Any unexpected changes, especially if an
1895
automated delivery sequence is used, require investigation.
1896
Rank
#199
RPN
64
Step#
161
Process
10. Plan
Prep
Step
5. Define imaging sequences to be used or
localization process
80
FM: Wrong imaging planned
1897
1898
Poor, incomplete, or wrong localization can be performed at treatment if the correct imaging
1899
sequences are not planned into the delivery process. This is an important issue for any IGRT
1900
delivery strategies. Progenitor causes include lack of standardized procedures, human failures,
1901
inadequate training, software error and inadequate programming. The failure mode is addressed
1902
through development of site-specific protocols and procedures and training of appropriate staff
1903
involved in the imaging process.
1904
Rank
#200
RPN
63
Step#
33
Process
5. Transfer images and
other DICOM data
FM: File corrupted
Step
Transfer primary (CT) dataset
1905
1906
This FM is unlikely with modern CT systems as they routinely provide reliable CT datasets and
1907
mainly fail during image acquisition. Systematic file corruption issues should be discoverable
1908
during comprehensive commissioning of the CT scanner and periodic QA as recommended by
1909
the AAPM TG-66 report. Random failures should be readily obvious through routine clinical
1910
practices and should be a part of comprehensive CT imaging QA program as recommended by
1911
the TG-66 report.
1912
Rank
#201
RPN Step#
Process
Step
62
16
3. CT/Sim
Volume needed for Tx Planning scanned
FM: 4. Patient “exceeds scan diameter” and volume of interest or volume
traversed by one or more beams is truncated.
1913
1914
A protocol for this situation should be defined. Attention and training of the simulator therapists
1915
will assist in preventing this problem.
1916
Rank
#202
RPN Step#
Process
Step
60
212
11. Day 1 Tx Tx delivered
FM: Some beam(s) delivered twice in one session
1917
1918
Addressed with rank 186 step 161.
1919
81
Rank
#203
RPN
59
Step#
160
Process
10. Plan
Prep
FM: Wrong imaging planned
Step
5. Define imaging sequences to be used for
localization process
1920
1921
Addressed with rank 199 step 161.
1922
Rank
#204
RPN
55
Step#
164
Process
10. Plan
Prep
FM: 1. Poor choice of sequence
Step
7. Decide delivery ordering of fields
1923
1924
See rank 172 step 132 for discussions.
1925
Rank
#205
RPN
53
Step#
143
Process
Step
9. Plan
2. Completion of formal prescription after
Approval
planning
FM: 6. Not signed when appropriate
1926
1927
Missing signatures and other forms of incomplete documentation can be included in the Physics
1928
pre-treatment review, weekly chart checks, and other checks performed to ensure billing
1929
compliance.
1930
Rank
#206
RPN
53
Step#
129
Process
Step
8. Tx
16. Evaluate delivery system limitations
Planning
FM: 1. Treatment volume inappropriate for delivery system
1931
1932
The initial plan directive and physician plan check should both prevent this error from
1933
propagating to treatment, though they will not prevent lots of extra work.
1934
Rank
#207
RPN
51
Step#
118
Process
8. Tx
Planning
FM: 2. Crash in sequencer
Step
10. Run leaf sequencer to create deliverable
plan
1935
82
1936
A crash in a program is not a problem unless bad files are written or a database entry is
1937
corrupted. See QM for sequencer issues in rank 54 step 120 and other entries.
1938
Rank
#208
RPN
49
Step#
131
Process
8. Tx
Planning
FM: 2. Undeliverable Tx plan
Step
15. Evaluate delivery system limitations
1939
1940
Addressed with rank 167 step 130.
1941
Rank
#209
RPN Step#
Process
Step
48
24
3. CT/Sim
CT images downloaded to next “station”
FM: Download not done in timely fashion
1942
1943
Inconvenient to patients. QM involves fixing process scheduling.
1944
Rank
#210
RPN
46
Step#
172
Process
Step
10. Plan
10. Download complete delivery plan to
Prep
delivery system
FM: 2. Incompatibility of plan data with delivery capabilities
1945
1946
This FM causes inconvenience to patients. QM for this kind of error involves fixing planning
1947
process to make sure capabilities are incorporated correctly.
1948
Rank
#211
RPN
44
Step#
36
Process
5. Transfer images
and other
DICOM data
FM: 2. File corrupted
Step
Transfer secondary (MRI, PET) datasets
1949
1950
See rank 200 step 33 for discussion.
1951
Rank
#212
RPN
35
Step#
28
Process
4. Other pre-Tx
imaging for CTV
localization
FM: a. Wrong area imaged
1952
83
Step
4. Scan performed
1953
This FM causes inconvenience to patients. QM for this kind of error involves fixing scheduling
1954
and information processes so that correct expectations are disseminated.
1955
Rank
#213
RPN
34
Step#
114
Process
8. Tx
planning
FM: 2. Transfer failure
Step
9. Transfer optimized fluence to leaf sequencer
(if separate system)
1956
1957
See rank 120 step 124 and rank 130 step 131 for discussion.
1958
Rank
#214
RPN
22
Step#
53
Process
7. RTP
Anatomy
FM: 3. File(s) corrupted
Step
Import images into RTP database
1959
1960
This interrupts the planning process but presents little hazard.
1961
Rank
#215
RPN
22
Step#
10
Process
2. Immobilization
and Positioning
FM: 2. Hardware lost, damaged
Step
Patient-specific hardware storage
1962
1963
This is inconvenient for patients. QM for this failure mode includes routine QA checks of
1964
integrity of all hardware used for treatment and adequate storage facilities.
1965
Rank
#216
RPN Step#
Process
Step
19
23
3. CT/Sim
Image set saved or sent to Tx planning
FM: Simulation image set accidentally deleted
1966
1967
This causes inconvenience to the patient. Quality management for this failure mode includes
1968
fixing the simulation process.
1969
84
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
REFERENCES
1
G.J. Kutcher, L. Coia, M. Gillin, W.F. Hanson, S. Leibel, R.J. Morton, J.R. Palta, J.A. Purdy,
L.E. Reinstein, G.K. Svensson, M. Weller, L. Wingfield, “Comprehensive QA for radiation
oncology: report of AAPM Radiation Therapy Committee Task Group 40,” Med. Phys. 21,
581-618 (1994).
2
E.E. Klein, J. Hanley, J. Bayouth, F.F. Yin, W. Simon, S. Dresser, C. Serago, F. Aguirre, L. Ma,
B. Arjomandy, and C. Liu, “Task Group 142 report: Quality assurance of medical accelerators,”
Med. Phys. 35, 4197–4212 (2009).
American Association of Physicists in Medicine Task Group Report 13, “Physical aspects of
quality assurance in radiotherapy” American Institute of Physics, New York, 1984.
3
4
P. B. Dunscombe, S. Iftody, N. Ploquin, E. U. Ekaette, and R. C. Lee, "The Equivalent Uniform
Dose as a severity metric for radiation treatment incidents," Radiother Oncol 84 (1), 64-6 (2007).
5
G. Yan, C. Liu, T. A. Simon, L. C. Peng, C. Fox, and J. G. Li, "On the sensitivity of patientspecific IMRT QA to MLC positioning errors," J Appl Clin Med Phys 10 (1), 2915 (2009)
6
T. Pawlicki, S. Yoo, L.E. Court, S. K. McMilan, R. K. Rice, J. D. Russell, and J. M. Pacyniak et
al., “Moving from IMRT QA measurements toward independent computer calculations using
control charts,” Radiother. Oncol.. 89, 330-337 (2008).
T. Pawlicki, M. Whitaker, and A.L. Boyer, “Statistical process control for radiotherapy quality
assurance,” Med. Phys. 32, 2777-2786 (2005).
7
8
M. Kapanen, M. Tenhunen, R. Parkkinen, P. Sipila, and H. Jarvinen, "The influence of output
measurement time interval and tolerance on treatment dose deviation in photon external beam
radiotherapy," Phys Med Biol 51 (19), 4857-67 (2006)
9
C. Constantinou and E. S. Sternick, "Reduction Of The Horns Observed On The Beam Profiles
Of A 6-Mv Linear-Accelerator," Medical Physics 11 (6), 840-842 (1984)
10
T. LoSasso, "IMRT delivery system QA," in Intensity modulated radiation therapy: the state of
the art. Medial Physics Monograph Vol. No. 29. Colorado Springs, edited by J. Palta and T. R.
Mackie (Medial Physics Publishing, Madison, WI, 2003), pp. 561–591]
11
A. Rangel and P. Dunscombe, "Tolerances on MLC leaf position accuracy for IMRT delivery
with a dynamic MLC," Medical Physics 36 (7), 3304-3309 (2009)
G Mu, E Ludlum, P Xia, “Impact of MLC leaf position errors on simple and complex IMRT
plans for head and neck cancer.” PMB 53, 77-88 (2008).
12
85
D A Low, JM Moran, et al “Dosimetry tools and techniques for IMRT”, Med Phys 38, 13131338 (2011)
14
T LoSasso “IMRT Delivery Performance with a Varian Multileaf Collimator”, IJROBP 71S,
S85-S88, 2008
2014
2015
2016
2017
2018
2019
13
2020
Approches for Intensity-Modulated Radiotherapy”, IJROBP 71S S93-S97, 2oo8
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
J E Bayouth, “Siemens Multileaf Collimator Characterization and Quality Assurance
15
16
C Liu, Simon TA, Fox C, et al, “Multileaf collimator characteristics and reliability
requirements for IMRT Elekta System”, IJROBP 71S, S 89-S92, 2008.
17
W. R. Lutz, R. D. Larsen, and B. E. Bjarngard, "Beam Alignment Tests For Therapy
Accelerators," Int Journal Of Radiat Oncol Biol Phys 7 (12), 1727-1731 (1981)
BF Fraass , K Doppke, M Hunt, et al, “American Association of Physicists in Medicine
Radiation Therapy Committee Task Group 53: Quality assurance for clinical radiotherapy
treatment planning.” Med Phys 25, 1773-1829 (1998).
18.
F Azmandian, D Kaeli, JG Dy et al, “Towards the development of an error checker for
2031
19
2032
radiotherapy treatment plans: a preliminary study.” PMB 52, 6511-24 (2007)
2033
E E Furhang, J Dolan, JK Sillanpaa, LB Harrison, “Automating the initial physics chart
2034
20
2035
checking process.” JACMP 10, 2855 (2009).
2036
R A Siochi, EC Pennington, TJ Waldron, JE Bayouth, “Radiation therapy plan checks in a
2037
21
2038
paperless clinic.” JACMP 10, 2905 (2009)
2039
22
2040
Treatment of Cancer, IAEA Technical Reports Series No. 430; http://www-
2041
pub.iaea.org/mtcd/publications/pdf/trs430_web.pdf
Commissioning and Quality Assurance of Computerized Planning Systems for Radiation
2042
2043
2044
2045
2046
2047
2048
2049
2050
23
Thomadsen B, Lin S-W, Laemmrich P, Waller T, Cheng A, Caldwell B, Rankin R, Stitt, J. Analysis of
Treatment Delivery Errors in Brachytherapy Using Formal Risk Analysis Techniques. Int J Radiat Oncol
Biol Phys 57: 1492 – 1508 (2003)
Siochi RA, Balter P, Block CD, et al, “A rapid communication from the AAPM Task Group
201: recommendations for the QA of external beam radiotherapy data transfer. AAPM TG201:
quality assurance of external beam radiotherapy data transfer”, JACMP 12, 3479 (2010).
24
86
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
25
P. W. McLaughlin, C. Evans, M. Feng, and V. Narayana, "Radiographic and Anatomic Basis
for Prostate Contouring Errors and Methods to Improve Prostate Contouring Accuracy," Int J
Radiat Oncol Biol Phys 76,369-78 (2010)
26
GA Ezzell, JM Galvin, D Low, et al. Guidance document on delivery, treatment planning, and
clinical implementation of IMRT: Report of the IMRT subcommittee of the AAPM radiation
therapy committee. Med. Phys. 30: 2089 – 2115 (2003);
G A Ezzell, JW Burmeister, N Dogan et al, “IMRT commissioning: Multiple institution
planning and dosimetry comparisons, a report from AAPM Task Group 119.” Med Phys . 36:
5359-5373 (2009)
27
Papanikolaou N, Battista JJ, Boyer AL, et al, “Tissue inhomogeneity corrections for
megavoltage photon beams, AAPM Report no. 85”,
http://www.aapm.org/pubs/reports/RPT_85.pdf
28
29
An investigation of the Therac-25 Accidents, N Leveson, CS Turner
http://courses.cs.vt.edu/cs3604/lib/Therac_25/Therac_1.html
30
The NY Times, NY, NY, Sunday January 24, 2010
(http://www.nytimes.com/2010/01/24/health/24radiation.html?pagewanted=all&_r=0)
Mutic S, Palta JR, Butker EK, et al, “Quality assurance for computed tomography simulators
and the computed tomography-simulation process: Report of the AAPM Radiation Therapy
Committee Task Group No. 66.”, Med Phys 30: 2762-2792 (2003).
31
EC Ford, K Smith, K Harris, S Terezakis, “Prevention of a wrong-location misadministration
through the use of an intradepartmental incident learning system.”, Med Phys 39: 6069-6971
(2012)
32
87