Development of a software based on an anthropomorphic
image atlas for image quality optimisation in interventional
radiology: initial results for 3D angiography
Poster No.:
C-3085
Congress:
ECR 2017
Type:
Scientific Exhibit
Authors:
F. Gardavaud , M. Gautier , H. Pasquier , F. Boudghene ; Paris/
1
1
2
1 1
2
FR, Creteil/FR
Keywords:
Radioprotection / Radiation dose, Interventional vascular,
Radiation physics, Cone beam CT, Dosimetry, Radiation safety,
Physics, Dosimetric comparison
DOI:
10.1594/ecr2017/C-3085
Any information contained in this pdf file is automatically generated from digital material
submitted to EPOS by third parties in the form of scientific presentations. References
to any names, marks, products, or services of third parties or hypertext links to thirdparty sites or information are provided solely as a convenience to you and do not in
any way constitute or imply ECR's endorsement, sponsorship or recommendation of the
third party, information, product or service. ECR is not responsible for the content of
these pages and does not make any representations regarding the content or accuracy
of material in this file.
As per copyright regulations, any unauthorised use of the material or parts thereof as
well as commercial reproduction or multiple distribution by any traditional or electronically
based reproduction/publication method ist strictly prohibited.
You agree to defend, indemnify, and hold ECR harmless from and against any and all
claims, damages, costs, and expenses, including attorneys' fees, arising from or related
to your use of these pages.
Please note: Links to movies, ppt slideshows and any other multimedia files are not
available in the pdf version of presentations.
www.myESR.org
Page 1 of 14
Aims and objectives
Context :
Currently, in order to optimize Image quality (IQ) in Interventional Radiology (IR), medical
physicists need :
•
•
Access to acquisition console in order to design/extract protocols
parameters [1,2]
To standardize protocols parameters in order to improve patient dose
management [3]
Radiologist validation to confirm IQ improvement
•
•
•
•
•
No guidelines or standards for protocol designing currently available
clinical routine can not allow installation full-time access
To much parameters to handle manually
Hard to quantify IQ without tailoring metrics
No extraction feature in DoseWatch
•
But :
For each clinical protocol, multiple parameters provide different IQ and Dose couple.
For example, in Digital Subtraction Angiography (DSA) mode, over 500 000 possibilities
could be reach (fig. 1).
Purpose :
To provide a protocol optimization software helping radiologist to tailor IQ in IR thanks
to an anthropomorphic images atlas.
Images for this section:
Page 2 of 14
Fig. 1: Parameters for angiographic protocol (without fluoro part)
© Service d'Imagerie Médicale, CHU Tenon - Paris/FR
Page 3 of 14
Methods and materials
Protocols database design and quality image levels
For the five most representative anatomic areas of our clinical practice (lung, rachis,
stomach, liver, pelvis), five reference protocols were set-up in our interventional system
(Innova IGS 540, GE Healthcare). Those protocols were defined with five different IQ/
dose preferences from low dose to high IQ (Fig. 2).
These preferences were defined by adjusting the Automatic Dose Rate Image Quality
(ADRIQ) controls logic strategies and the detail levels for 3D Angiography (3DA) (fig. 3)
and fluoro modes (fig. 4). The clinical routine protocols used in our department were GE
Standard Reference and GE Low Dose Reference.
Acquisitions on an anthropomorphic phantom
For each reference protocol and each IQ level, 3DA images were acquired on an
anthropomorphic phantom (PBU-60®, KYOTO KAGAKU, fig. 5). The same process was
realized for fluoro acquisitions with five sequences for each IQ levels to evaluate 0 0 1 2
12 Hopital Tenon 1 1 13 14.0 96 Normal 0 21 false false false FR X-NONE X-NONE
Peak Dose Skin (PDS) and Peak Dose Skin Rate (PDSR) for 3DA and fluoro
mode respectively were calculated for each reference protocols by using DICOM
objects information [4,5,6] and Em.Dose (Esprimed) solutions. Film dosimetry solution
(GafChromic® films) was not used for this study because this dosimetry way was very
dependent on energy beam and ADRIQ strategies, which tailored energy beam, were
employed [7].
Beforehand, the medical physicist controlled the IR installation to ensure that the
displayed FOV, SOD and Frame rate were physically robust.
A senior radiologist qualitatively validated the anthropomorphic images of the reference
protocols offered in ProtoEnhanceIR.
Software for IR optimization: ProtoEnhanceIR
Page 4 of 14
The authors developed a software, named ProtoEnhanceIR, to help to optimize IR
protocols in 3DA and fluoro mode by displaying the anthropomorphic images with the
associated PDSR or PDS. The software was coded in Python language and can be used
in the most common operating systems (Windows®, OS X® and most common Linux
distributions).
The user imports his own protocol database from an USB device. Then, ProtoEnhanceIR
recognizes all the user protocols and proposes to select a protocol to optimize. The
user has to match the chosen protocol with the corresponding reference protocol. Then,
ProtoEnhanceIR displays the five IQ levels sequences with the associated PDSR or PDS.
ProtoEnhanceIR includes a DICOM viewer allowing to scroll through the axial images or
2D images and to select a windowing. Also, ProtoEnhanceIR offers to view the resulting
dose map, extracted from Em.Dose solution, for each reference protocols (fig. 6)
Thus, Radiologists are able to qualify as a routine examination, the five proposed
optimizations and make a choice according to their clinical needs. Once the user selected
the IQ level he preferred, ProtoEnhanceIR generates IR system files that are semiautomatically implemented from the USB device into the acquisition console without
generating human errors due to manual data entry. The computing specific workflow was
illustrated fig. 7.
The whole material&method process was illustrated fig.8.
Images for this section:
Fig. 2: Protocols design tree. For the five clinical protocols, five protocols with different
IQ are proposed. 25 protocols in totality are available for the radiologist.
Page 5 of 14
© Service d'Imagerie Médicale, CHU Tenon - Paris/FR
Fig. 3: 3DA protocols parameters to design the five IQ levels.
© Service d'Imagerie Médicale, CHU Tenon - Paris/FR
Fig. 4: Fluoro protocols parameters to design the five IQ levels.
© Service d'Imagerie Médicale, CHU Tenon - Paris/FR
Page 6 of 14
Fig. 5: Photography of the PBU-60 anthropomorphic phantom.
© KOYTO KAGAKU
Page 7 of 14
Fig. 6: Example of a dose map extracted from Em.Dose solution. Em.Dose works even
in 3DA acquisition mode.
© Service d'Imagerie Médicale, CHU Tenon - Paris/FR
Page 8 of 14
Fig. 7: ProtoEnhanceIR computing workflow
© Service d'Imagerie Médicale, CHU Tenon - Paris/FR
Page 9 of 14
Fig. 8: material&method process. reference protocols were designed. for each IQ;
sequences were acquired in 3DA and fluoro mode on an anthropomorphic phantom; a
software was design to provide images atlas and to tailor clinical protocols; system files
were set-up to be implemented on Innova system.
© Service d'Imagerie Médicale, CHU Tenon - Paris/FR
Page 10 of 14
Results
Evaluation of 3DA acquisition protocols between clinical routine protocols and
ProtoEnhanceIR protocols on the anthropomorphic phantom
As shown in fig. 9, compared to our 3DA clinical routine low dose protocol (GE Low Dose
Reference), the 3DA optimized protocol proposed by ProtoEnhanceIR, named Ultra Low
Dose, was really similar in term of PDS. In fact, in 3DA mode, RDLplus and RDLstandard
for ADRIQ strategy had the same performance. So, it yieled to very similar PSD values.
Indeed, for lung, rachis, stomach, liver, pelvis PSDUltra_Low_Dose was 4.73; 21.51; 16.1;
15.69; 15.09 mGy while PSDGE_Low_Dose_Reference was 4.74; 21.67; 16.1; 15.56; 14.98
mGy respectively.
The same situation was observed between the two following IQ levels : GE Standard
reference and High Image quality. Also, in 3DA mode, IQplus and IQstandard for ADRIQ
strategy had the same performance. So, it yielded to very similar PSD values.
However, compared to our 3DA clinical routine standard protocol (GE Standard
Reference), the 3DA optimized protocol proposed by ProtoEnhanceIR, named Standard,
allowed an average dose reduction by a mean factor in the five anatomic regions of
1,9. Indeed, for lung, rachis, stomach, liver, pelvis, PSDStandard was 7.74; 35.88; 28.61;
27.91; 29.55 mGy while PSDGE_Standard_Reference was 13.4; 74.18; 54.99; 53.79; 53.08
mGy respectively.
Evaluation of fluoro acquisition protocols between clinical routine protocols and
ProtoEnhanceIR protocols on the anthropomorphic phantom
As shown in fig. 10, compared to our fluoro clinical routine protocol, the fluoro optimized
protocols proposed by ProtoEnhanceIR, named Low Dose and Ultra Low Dose, allowed
an average dose reduction by a mean factor in the five anatomic regions of 1,8 and 3,8 for
Low Dose and Ultra Low Dose IQ levels respectively with qualitatively validated clinical
IQ. Indeed, for lung, rachis, stomach, liver, pelvis, PSDRLow_Dose was 4.6 ± 0.15; 64.58
-1
± 2.17; 20.09 ± 0.67; 20.89 ± 0.7; 23.53 ± 0.79 µGy.s while PSDRGE_Low_Dose_Reference
-1
was 9.12 ± 0.3; 92.94 ± 3.13; 39.99 ± 1.34; 39.73 ± 1.33; 39.74 ± 1.33 µGy.s
respectively. Also, for lung, rachis, stomach, liver, pelvis, PSDRUltra_Low_Dose was 1.88 ±
-1
0.06; 29.88 ± 1.0; 10.40 ± 0.35; 11.2 ± 0.37; 10.87 ± 0.36 µGy.s .
Page 11 of 14
Images for this section:
Fig. 9: Peak skin dose for the five anatomic regions and for each IQ levels. GE low
Dose Reference and GE Standard Reference was the two clinical routine IQ levels in
our department.
© Service d'Imagerie Médicale, CHU Tenon - Paris/FR
Fig. 10: Peak skin dose rate for the five anatomic regions and for each IQ levels. GE
low Dose Reference and GE Standard Reference was the two clinical routine IQ levels
in our department.
© Service d'Imagerie Médicale, CHU Tenon - Paris/FR
Page 12 of 14
Conclusion
In this study, we developed a software, named ProtoEnhanceIR, to help radiologists
and medical physicists to create as well as possible, optimized IR protocols for patients.
ProtoEnhanceIR offered the possibility to prospectively design, without any patient
exposition but only on an anthropomorphic phantom, new protocols. Indeed, radiologists
could evaluate the resulting protocols IQ and radiation exposure and could achieve
their needs in term of patient care. Moreover, radiologists and medical physicists no
longer needed to use the console acquisition because ProtoEnhanceIR offered direct
protocol implementation also removing human input errors. They could save their patients
examinations number while leading an optimization policy thanks to ProtoEnhanceIR.
From our clinical experience, ProtoEnhanceIR offers significant dose reduction
opportunities while maintaining clinical IQ.
Limitations
There are some limitations on this study. First, ProtoEnhanceIR was not a simulator and
users had to match their IR protocols with ProtoEnhanceIR protocols database. However,
we built ProtoEnhanceIR with an adaptive structure to easily implement additional IR
protocols to fit with user's clinical practice. Second, ProtoEnhanceIR allows user to screen
through images in the only axial plane and have also a cine mode for fluoro acquisition.
Last, in this study, anthropomorphic phantom IQ was not compared to patient IQ. Nearly,
we prospect to evaluate phantom IQ.
As far as we know, there is no similar solution to drive radiologists and medical physicists
in their optimization.
Personal information
References
[1] Donald L. Miller et al. Quality Improvement Guidelines for Recording Patient Radiation
Dose in the Medical Record. J Vasc Interv Radiol 2009; 20:S200-S207
Page 13 of 14
[2] Donald L. Miller et al. Quality Improvement Guidelines for Recording Patient Radiation
Dose in the Medical Record for Fluoroscopically Guided Procedures. J Vasc Interv Radiol
2012; 23:11-18
[3] Gabriel Bartal et al. Management of Patient and Staff Radiation Dose in Interventional
Radiology: Current Concepts. Cardiovasc Intervent Radiol (2014) 37:289-298
[4] A. Kyle Jones, Joe E. Ensor, and Alexander S. Pasciak. How accurately can the peak
skin dose in fluoroscopy be determined using indirect dose metrics ?. Medical Physics
41, 071913 (2014); doi: 10.1118/1.4884020
[5] A. Kyle Jones, Alexander S. Pasciak. Calculating the peak skin dose resulting from
fluoroscopically guided interventions. Part I: Methods. Journal of Applied Clinical Medical
Physics, Vol 12, No 4 (2011).
[6] A. Kyle Jones, Alexander S. Pasciak. Calculating the peak skin dose resulting from
uoroscopically-guided interventions. Part II: Case studies. JournAL oF APPLIEd cLInIcAL
MEdIcAL PHYSIcS, VoLuME 13, nuMBEr 1, 2012
[7] J. Farah et al. Characterization of XR-RV3 GafChromic® films in standard laboratory
and in clinical conditions and means to evaluate uncertainties and reduce errors. Medical
Physics 42, 4211 (2015); doi: 10.1118/1.4922132
Page 14 of 14