British Journal of Rheumatology 1996;35(snppl. 3): 18-22
MICROFOCAL TECHNIQUES IN QUANTITATIVE RADIOGRAPHY:
MEASUREMENT OF CANCELLOUS BONE ORGANIZATION
J. C BUCKLAND-WRIGHT, J. A. LYNCH and C BIRD
Division ofAnatomy andCellBiology. United Medicaland DentalSchools ofGuy's andSt Thomas's Hospitals,
London University, London
SUMMARY
Microfocal radiography records, with unusually good resolution, the detailed structural organization of cancellous bone. A textural
imaging method, fractal signature analysis (FSA), was used to quantify the horizontal and vertical trabecular organization recorded
within macroradiographic images of the spine of post-menopausal women and the tibia in osteoarthridc knees, and the analysis of
variance method was applied to the wrist and hand of rheumatoid patients. Changes in trabecular structure were found to correlate
with (i) body weight, age and bone mineral density in the lumbar spine of post-menopausal women; (ii) the degree of cartilage loss
and age in the tibia of patients with knee OA; and (iii) analysis of variance quantified the extent of 'normal', osteopaenic and
eroded bone in rheumatoid joints. Quantitation of cancellous bone organization can add significantly to our understanding of
disease processes and effect of therapy in diseased joints.
KEY WORDS:
Cancellous bone, Fractal signature analysis, Microfocal radiography, Osteoarthritis, Osteoporosis, Rheumatoid
arthritis.
QUANTITATIVE MICROFOCAL RADIOGRAPHY
tubes are characterized by a micronsized X-ray source in which the object being examined is
placed close to the X-ray tube and the film is positioned
1.5-2 m away, resulting in magnification (macroradiographs) of the image with high spatial resolution [1, 2].
For accurate and reproducible measurements of the
features recorded in macroradiographs, standardized
procedures are employed to define the precise
anatomical position of the joint for radiography [3-6];
to correct for the effect of radiographic magnification
[3—6]; to define precisely the anatomical boundaries
used for measurements of features [3-6]; and to
determine the reproducibility of the measurement from
an assessment of the coefficient of variation for repeat
measures and test-retest procedures [6-12].
Quantitative microfocal radiographic examination of
arthritic patients allows measurement of the extent [13,
14] and progression [IS] of disease in the rheumatoid
hand from changes in erosion size. Assessment of the
effects of therapy in patients with early rheumatoid
arthritis, treated with Myocrisin, showed not only a halt
in erosion area progression, but, importantly, erosion
repair [10]. In patients with knee OA, quantifying the
extent and progression of the radiographic features
revealed that bony features, osteophytosis and
subchondral sclerosis, preceded the onset of cartilage
loss measured as joint space narrowing [11]. Precise
methods for measuring joint space width permitted the
detection of the effects of non-steroidal antiinflammatory therapy on cartilage in knee OA [16].
ability to record, with unusually good resolution, the
fine detail of cancellous bone The difference between
compact and cancellous bone is that the latter has a
greater surface area to volume ratio available to cellular
activity and greater tissue vascularity. Due to the rapid
rate of trabecular bone turnover [17], its response to
unloading, increased blood flow and synovitis [18],
cancellous bone is more sensitive to changes to
physiological and biomechanical environment than
compact bone.
MICROFOCAL X-ray
QUANTIFYING CANCELLOUS BONE
ORGANIZATION
We report the results of a new method for quantifying
the trabecular organization using a recently developed
method of fractal signature analysis (FSA) [19, 20].
Fractal analysis measures the degree of 'roughness' of
an image of those structures, and also quantifies the
change in 'roughness' with alterations in spatial scale
[21]. Self-similar images (looking the same at all
magnifications) [22] are said to be 'fractal' and have
associated with them a fractal dimension, which is
between 2 and 3 for a surface [21]. When the pattern of
a structure has altered at a particular size or sizes so as
to be no longer self-similar, the 'fractal signature' of its
image quantifies the alteration in the fractal dimension
of the structure, and the size(s) at which those changes
have occurred [19, 20]. The fractal dimension, and
similarly the fractal signature, has no units since it is
calculated from the ratio of two areas [21]. The fractal
dimension of cancellous bone assesses the composite
nature of the tissue, which is determined principally by
trabecular number, spacing and cross-connectivity [23],
features which are here referred to as trabecular
structures. FSA provides more information than the
mean fractal dimension employed by previous
investigators [20, 21] which only quantifies the overall
CANCELLOUS BONE
A major advantage of microfocal radiography is its
Correspondence to: Dr J. C. Buckland-Wright, Division of
Anatomy, UMDS Guy1* Hospital, London Bridge, London SE1 9RT.
O 1996 British Society for Rheumatology
18
BUCKLAND-WRIGHT ETAL.: QUANTITATIVE MICROFOCAL RADIOGRAPHY
3.2
o 3.1
3.0
2.9
3.1
>
3.0
2.9
2.8
2.8
2.7
5 2.7
-3
Ctl
I
3.2
3 rtic
CD
ision
i
2.6
£ 2.5
0.24
0.48
0.72
0.96
Size of trabecular structures (mm)
2.6
2.5
1.20
19
•
*
0.24
0.48
0.72
0.96
1.20
Size of trabecular structures (mm)
B
FlO. 1.—(A) The mean (IB.) fractal signature for horizontal trabecular structures was similar in early postmenopausal women with low ( • ) and high
( • ) bone mineral Hwiriti^ (B) The mean (&£.) fractal signature for vertical trabecular structures was greater in early postmenopausal women with
low ( • ) compared with those with high ( • ) bone mineral rkm^tvm The significance of difference in the vertical structures between the two groups is
indicated by the symbols under the signature for the high BMD group: » = P< 0.0025, * = P < 0.025, +• = P < 0.04.
appearance of cancellous bone, whereas FSA measures
separately the fractal dimension of the horizontal and
vertical trabecular structures within the cancellous bone
[19, 20] and quantifies how the fractal dimension varies
with the size of the structures.
CANCELLOUS BONE CHANGES IN THE SPINE
OF POST-MENOPAUSAL WOMEN
The trabecular organization in the lumbar spine of
post-menopausal women was obtained using FSA, and
its relationship to vertebral bone mineral density
(BMD), patient's age, weight and body mass index
determined [24].
BMD measurements, obtained by dual-energy X-ray
absorptiometry (DXA) using the Hologic QDA-1000,
and X4 macroradiographs, were taken of the lumbar
spine (L1-L4) from 47 early post-menopausal (3-36
months) women (mean age 54.1 ± 3.2 yr). Patients were
divided into two groups on the basis of their BMD:
those with low BMD (mean 0.86 ± 0.17 g/cm2, n = 19)
and high BMD (mean 1.06 ± 0.12 g/cm2, n = 28).
The results showed that (i) patient weight and body
mass index correlated with the fractal signature values
for thick vertical trabecular structures ranging in size
from 0.5 to 1.0 mm (P < 0.005-0.05); (ii) age correlated
with fractal signature values for fine horizontal
structures (size range 0.1-0.25 mm) (P < 0.005-0.05);
BMD correlated with fractal signature values for fine
vertical structures (size range 0.1-0.4 mm) (P <
0.005-0.05) [24]. The low BMD patients also had a
significantly raised fractal signature for fine vertical
structures (size range 0.18-0.84 mm) when compared to
the patients with high BMD (P < 0.003-0.04) [24]
(Fig. 1).
The correlation of FSA for thick vertical trabecular
structures with weight confirms the expected association between the thicker weight-bearing trabeculae and
load transmission. The correlation between age and
FSA for fine trabecular structures indicates that FSA
quantifies age-related changes, which include a
reduction in the number of horizontal trabeculae [25].
FSA of the fine vertical trabecular structures correlated
with reduced BMD and appears to correspond to early
stages of axial bone loss in post-menopausal women. In
this process, focal perforation of coarse vertical
trabecular plates progresses into a lattice of bars and
rods [25]. These changes would result in an increase in
the number and cross-connectivity of fine trabecular
structures and an increase in the 'roughness' in the
radiographic appearance of the cancellous bone organization, leading to higher fractal signature values [24].
CANCELLOUS BONE CHANGES IN KNEE
OSTEOARTHRITIS
FSA was used to quantify tibial cancellous bone
changes in osteoarthritic knees with minimal and
definite joint space narrowing (JSN) compared to knees
without arthritis [26].
Thirty-three patients with knee OA (9 male, 24
female, mean age 65.5 ± 10.2 yr) were selected on the
basis of clinical and radiographic evidence of OA. X5
macroradiographs were taken of both knees in the
standing semi-flexed and weight-bearing tunnel views
[11]. Macroradiographs of the knees of 14 non-OA
subjects (mean age 37.0 ± 10.6 yr) were also taken. Joint
space width (JSW) was measured in the medial and
lateral compartments using a computerized technique
[12]. A box analysis was set up based on Ahlbacks'
criteria [27], and OA knees were grouped on their
minimum JSW in the medial compartment, into those
with early OA (JSW > 3 mm) and those with definite
OA (JSW < 3 mm) in either or both views. Fractal
signatures for horizontal and vertical trabecular
structures (size range 0.06-1.20 mm) were calculated for
the central region of subchondral cancellous bone in
each tibial compartment of all knees. The differences in
the fractal signatures between groups of knees were
examined for significance using multivariate analysis of
variance (MANOVA) with repeated measures.
All OA knees had medial compartment disease. FSA
of the trabecular structure of the medial diseased
compartment of the tibia was significantly different
MRI IN THE ASSESSMENT OF RA
20
TABLEI
Using mean image variance, regions of 'normal', osteopacnic and eroded bone were found to be statistically significantly different between each of
the categories as shown in (A); the ability of this image texture measure to identify correctly the different categories of bone is shown in (B)
(A) Mean image variance vs class of cancellous bone
Bone type
Normal
Eroded
Osteopaenic
Variance of intensity1
16.13(10.68)
6.88 (4.48)
9.60 (7.70)
Significance between groups'*
rs normal bone
vs eroded bone
P< 0.00001
P < 0.00001
P < 0.0053
(B) Image texture as a classifier of cancellous bone
Predicted bone group (%)
Actual bone group
Normal
Osteopaenic
Eroded
Normal
Osteopaenic
Eroded
61.3
21.0
0.0
29.0
46.8
14.5
9.7
3Z3
85.5
"Values are mean (six).
b
Wilcoxon matched pairs test
(P< 0.003) from that in the non-arthritic joints:
changes indicated that horizontal trabeculae structures
(size range 0.36-0.66 mm) increased in thickness
significantly in knees with early OA (joint space > 3
mm) (P < 0.001) and that vertical trabecular structures
(size -0.77 mm) increased in knees with marked JSN
(joint space < 3 mm) in both views and showed a
significant increase in those knees with definite JSN in
the tunnel view only (P < 0.001). In the lateral compartment of the tibia, FSA did not show a difference
between any of the categories. With increasing age of all
subjects, FSA showed a significant increase in the
number of fine horizontal and vertical trabeculae
structures (size range 0.18-0.30 mm) (P < 0.005-0.02).
No correlation was found between the subjects' body
weight and changes in the subarticular cancellous bone
organization.
In the medial diseased compartment of the tibia in
OA knees, the horizontal trabeculae increased in thickness in early OA and vertical trabecoilar connectivity
increased in knees with more advanced disease [26].
With increasing age of the subject, a rise in the number
of fine trabecula structures occurred in the proximal
end of the tibia.
CANCELLOUS BONE CHANGES IN
RHEUMATOID HANDS
Two types of bony change are recorded radiographically in rheumatoid arthritis: erosions and juxtaarticular osteopaenia. Current methods of assessment
use changes in the numbers of the first, and tend to
neglect the second. An image texture analysis method
[28] was used to quantify differences in the cancellous
bone between region of 'normal', eroded and osteopaenic bone.
This textural method is based upon fact that an image
with rapid and repetitive changes in intensity (fine
texture) appears different to one with slower changes
(coarse texture). Thus, the intensity of a fine-textured
image will vary more than a coarse-textured image in
any unit area, given that both images have the same
mean intensity. As a result, measures of statistical
spread such as variance will have higher values for
fine-textured images [28]. Such measurements can be
made in different directions, and will have markedly
different values for cancellous bone with different
structural appearance.
X5 macroradiographs were taken of the hands of 10
rheumatoid patients. A total of 62 images of metacarpophalangeal, proximal and distal interphalangeal
joints were assessed. The variance of intensity was
measured for regions of 'normal', osteopaenic and
eroded bone (Table I). A grid was imposed upon each
X-ray image. The images within each square of the grid,
or sub-images, were individually analysed and the
values used to produce a map of the different types of
bone: 'normal', osteopaenic and eroded (Fig. 2).
Significant differences in image texture were found
between the three types of bone (Table I) recorded in the
macroradiographs. The significantly lower variances for
osteopaenic and eroded bone, compared to 'normal',
are attributable to the loss of trabeculae, resulting in a
more homogeneous image (Table LA). The progressive
changes that occur in the osteopaenic bone represent
the continuous nature of the inflammatory process,
resulting in an overlap at both the 'normal' and eroded
extremes in the values for osteopaenic bone (Table IB).
Simple texture measures show promise in automatically
quantifying the extent of both erosions and osteopaenic
bone.
DISCUSSION
The methods for clinically examining cancellous bone
in vivo do not directly measure the detailed structure at
the lumbar spine, knee joint and wrist. DXA examines
mineral content, but does not quantify structure.
BUCKLAND-WRIGHT ETAL: QUANTITATIVE MICROFOCAL RADIOGRAPHY
21
studies of post-mortem tissue, demonstrating the
capacity of this technique to quantify those trabecular
features directly related to its strength, i.e. anisotropy
and connectivity [34].
ACKNOWLEDGEMENTS
We would like to thank The Wellcome Trust for
supporting the work undertaken by Dr Lynch, and the
Arthritis and Rheumatism Council for supporting Dr
Bird's work in rheumatoid arthritis.
REFERENCES
Fio. 2.—(Top) Part of a macroradiograph of the wrist showing a large
erosion at the distal end of the radius (radiographic magnification *5,
reproduced at x2.1). (Bottom) Processed image of the same
macroradiograph showing the different types of bone obtained from
the ratio of the variances in the x and y directions within each square
of the grid or sub-image. Normal bone is bright and osteopaenic and
eroded bone light to dark grey.
Quantitative computed tomography (QCT) measures
bone mineral and gross trabecular structure, but has a
resolution limitation of ~0.5 mm, and trabecular
structures, particularly in the spine, are at sizes below
this value [29]. MRI can indirectly detect changes in
trabeculae from magnetic susceptibility artefacts and
alterations in the parameter T2* [30], but fine trabecular
detail is lost since resolution is limited to ~0.5 mm.
Interpretation of these images is difficult and involves
the use of artefacts to examine changes occurring at
very fine resolution, which are not seen even in special
high-resolution scans [31]. The particular pulse
sequences used cause variations in the apparent size and
structure of the trabecular network [32]. Standard
radiographs can be scored for changes in trabecular
architecture [33], but this technique is insensitive for
quantifying subtle changes in bone structure. Whereas
image analysis techniques of FSA and variance
quantify cancellous bone organization in macroradiographs of (i) the lumbar spine of post-menopausal
women, leading to a better understanding of the
relationship of BMD and bone structure and their role
in predicting patients at risk of spinal fracture; (ii) the
tibia of patients with knee OA, providing an insight into
the relationship between subchondral bone changes and
articular cartilage loss; and (iii) the hands of rheumatoid patients, providing a far more sensitive measure of
disease progression and therapeutic response than either
erosion number or size. Recently, validation of FSA as a
method for quantifying cancellous bone structure,
recorded in macroradiographs, has come from detailed
1. Buckland-Wright JC. A new high definition microfocal
X-ray unit Br J Radiol 1989;62201-8.
2. Buckland-Wright JC, Bradshaw CR. Clinical applications
of high definition microfocal radiography. Br J Radiol
1989;62209-17.
3. Buckland-Wright J C Quantitative radiography of osteoarthritis. Ann Rheum Dis 1994;53:268-75.
4. Buckland-Wright JC, Macfarlane DG. Radioanatomic
assessment of therapeutic outcome in osteoarthritis. In
Kuettner KE, Goldberg V eds, New horizons in
osteoarthritis. Rosemont: American Academy of Orthopaedic Surgery, 1995:51-65.
5. Buckland-Wright C. Protocols for precise radio-anatomical positioning of the tibiofemoral and patellofemoral
compartments of the knee. Osteoarthritis Cartilage
1995;3(SnpplA):71-«0.
6. Buckland-Wright JC, Macfarlane DG, Ward RJ, Williams
SA. Accuracy and precision of joint space width
measurements in standard and macroradiographs of
osteoarthritic knees. Am Rheum Dis 1995;54:872-80.
7. Buckland-Wright J C Carmichael I, Walker SR. Quantitative microfocal radiography accurately detects joint
changes in rheumatoid arthritis. Ann Rheum Dis 1986: 45:
463-7.
8. Buckland-Wright JC, Macfarlane DG, Lynch JA, Clark B.
Quantitative microfocal radiographic assessment of
progression in osteoarthritis of the hand. Arthritis Rheum
1990; 33: 57-65.
9. Buckland-Wright JC, Macfarlane DG, Lynch JA.
Relationship between joint space width and subchondral
sclerosis in the osteoarthritic hand: a quantitative
microfocal study. / Rheumatol 1992;19:788-95.
10. Buckland-Wright J C Clarke GS, Chikanza IC, Grahame
R. Quantitative microfocal radiography detects changes in
erosion area in patients with early rheumatoid arthritis
treated with myocrisine. J Rheumatol 1993;20:243-7.
11 Buckland-Wright J C Macfarlane DG, Jasani MK, Lynch
JA. Quantitative microfocal radiographic assessment of
osteoarthritis of the knee from weight bearing tunnel and
semi-flexed standing views. J Rheumatol 1994^1:1734-41.
12. Lynch JA, Buckland-Wright JC, Macfarlane DG.
Precision of joint space width measurement in knee
osteoarthritis from digital image analysis of high
definition macroradiographs. Osteoarthritis Cartilage
1993;l:209-18.
13. Buckland-Wright JC. Microfocal radiographic examination of erosions in the wrist and hands of patients with
rheumatoid arthritis. Ann Rheum Dis 1984;43:160-71.
14. Buckland-Wright JC, Walker SR. Incidence and size of
erosions in the wrist and hand of rheumatoid patients: a
quantitative microfocal radiographic study. Arm Rheum
Dis 1987;46:463-7.
15. Buckland-Wright JC, Clarke GS, Walker SR. Erosion
number and area progression in the wrist and hands of
22
MRI IN THE ASSESSMENT OF RA
rheumatoid patients: a quantitative microfocal radiographic study. Ann Rheum Dis 1989;48:25-9
16. Buckland-Wright JC, Macfarlane DG, Lynch JA, Jasani
MK. Quantitative microfocal radiography detects changes
in OA knee joint space width in patients in placebocontrolled trial of NSAID therapy. / Rheumatol 1995;
22:937-43.
17. Gong JK, Burgess E, Bacalao P. Accretion and exchange
of strontium-85 in trabecular and cortical bone. Radiat
Res 1966;28:753-65.
18. Dedrick DK, Goldstein SA, Brandt KD, O'Connor BL,
Goulet RW, Albrecht MA. Longitudinal study of subchondral plate and trabecular bone in cruciate-deficient
dogs with osteoarthritis followed up for 54 months.
Arthritis Rheum 1993;36:1460-7.
19. Lynch JA, Hawkes DJ, Buckland-Wright JC Analysis of
texture in macroradiographs of osteoarthritic knees using
the fractal signature. Phys Med Biol 1991^6:709-22.
20. Lynch JA, Hawkes DJ, Buckland-Wright JC. A robust and
accurate method for calculating the fractal signature of
texture in macroradiographs of osteoarthritic knees. Med
Inform 1991 ;16:241-51.
21. Pentland AP. Fractal based descriptions of natural scenes.
IEEE Trans Pattern Ann Machine Intell PAMI 1984;
6:661-74.
22. Feder J. Fractals. New York: Plenum Press, 1988.
23. Majumdar S, Weistein RS, Prased RR, Genant HK. The
fractal dimension of trabecular bone: a measure of
trabecular structure. Calcif Tissue Int 1993^2:168.
24. Buckland-Wright JC, Lynch JA, Rymer J, Fogelman I.
Fractal signature analysis of macroradiographs measures
trabecular organization in lumbar vertebrae of postmenopausal women. Calcif Tissue Int 1994^4:106-12.
25. Parfitt AM. Trabecular bone architecture in the pathogenesis and prevention of fracture. Am J Med 1987;
82(Sappl IB): 68-72.
26. Lynch JA, Macfarlane DG, Buckland-Wright JC. Fractal
signature analysis measures trabecular organisation in
macroradiographs of OA knee patients. Trans Orthop Res
Soc 1995;20:304.
27. Ahlback S. Osteoarthritis of the knee: a radiographic
investigation. Ada Radiol 1968;277(SiippL):7-72.
28. Gonzalz RC, Woods RE. Digital image processing. New
York: McGraw-Hill, 1992:506-10.
29. Bergot C, Laval-Jeantet AM, Preteux F, Meunier A.
Measurement of anisotropic vertebral trabecular bone loss
during aging by quantitative image analysis. Calcif Tissue
Int 1988;43:143-9.
30. Wehrli FW, Ford JC, Artie M, Kressel HY, Kaplan FS.
Trabecular structure: preliminary application of MR
interferometry. Radiology 1991;179:615-21.
31. Jara H, Wehrli FW, Chung H, Ford JC. High-resolution
variable flip angle 3D MR imaging of trabecular structure
in vivo. Magn Reson Med 1993;29:528-39.
32. Jergas MD, Majumdar S, Keyak JH et al Relationships
between Young modulus of elasticity, ash density, and
MRI derived effective transverse relaxation T2* in tibial
specimens. / Comput Assist Tomogr 1996;19:472-9.
33. Singh M, Nagrath AR, Maini PS. Changes in trabecular
pattern of the upper end of the femur as an index of
osteoporosis. / Bone Joint Surg 1970;52A:457-467.
34. Lynch JA, Buckland-Wright JC, Hawes DJ, Nair SV.
Changes in anisotropy of modelled bone measured by
simulated radiography and fractal signature analysis.
Trans Orthop Res Soc 1996^21:714.