This Article Abstract Figures Only Full Text (PDF) Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Gandy, S J Articles by Watt, I Search for Related Content PubMed PubMed Citation Articles by Gandy, S J Articles by Watt, I

Measurement of cartilage volumes in rheumatoid arthritis using MRI

¹ Department of Medical Physics & Bioengineering, Bristol General Hospital, Guinea Street, Bristol BS1 6SY, ² Rheumatology Unit, University of Bristol Division of Medicine, Bristol Royal Infirmary, Bristol BS2 8HW, ³ Division of Imaging Sciences and Biomedical Imaging, University of Manchester, Stopford Building, Oxford Road, Manchester M13 9PT, ⁴ MRC Health Services Research, Canynge Hall, Whiteladies Road, Bristol BS8 2PR, ⁵ Respiratory and Inflammation Research, ⁶ Enabling Science Technology and Information, AstraZeneca, Alderley Park, Macclesfield, Cheshire SK10 4TG and ⁷ Department of Clinical Radiology, Bristol Royal Infirmary, Bristol BS2 8HW, UK

	Abstract

Top Abstract Introduction Method and materials Results Discussion Conclusion References

 
MRI is a valuable imaging modality for assessment of the articular cartilage in rheumatoid arthritis (RA) and is potentially of use in monitoring disease progression and response to therapy. In this study, we investigated the sources of error in volume measurements obtained by segmentation of MR images of knee cartilage in patients with RA and followed cartilage volume in a group of RA patients for 12 months. 23 RA patient volunteers were recruited for knee imaging. Six subjects were imaged at baseline only, six were imaged at baseline and again within an hour in the same imaging session, six subjects were imaged at baseline and 7 days, and 17 subjects were imaged at baseline, 4±2 months and 12 months. Imaging was performed at 1.0 T using a three-dimensional spoiled gradient-echo sequence with fat-suppression. Manual image segmentation was performed once or twice on the lateral tibial, medial tibial, patellar and femoral compartment by either one or two segmenters. Coefficients of variation (CoV) for repeated volume measurement of total cartilage were 2.2% (same segmenter, same scan), 5.2% (different segmenter, same scan), 4.9% (same segmenter, different scan, same session), and 4.4% (same segmenter, different scan, different session). Over the 12 month duration of the study there was no significant change in total cartilage volume, nor were there significant changes in volume in any individual compartment. This measurement technique is reproducible, but any net change in cartilage volume over 1 year is very small.

	Introduction

Top Abstract Introduction Method and materials Results Discussion Conclusion References

 
Rheumatoid arthritis (RA) is characterized by a chronic proliferative synovitis, or pannus, which leads to bone erosions and destruction of the underlying articular hyaline cartilage. Research on strategies to prevent destruction of the articular cartilage in arthritis requires reliable measurement of the clinical effectiveness of intervention. Disease activity in RA has been traditionally investigated by monitoring clinical indices of disease such as C-reactive protein (CRP) or erthyrocyte sedimentation rate (ESR), together with global assessment of joint pain and disability scores (e.g. Health assessment questionnaire (HAQ)) and characterization of damage to the tissues by radiographic scoring using Larsen or Sharp scores [1, 2]. The current gold-standard for in vivo assessment of change to hyaline cartilage is X-radiographic measurement of the joint space or interbone distance [3] but this technique provides only an indirect assessment of cartilage and lacks precision.

MRI is potentially of value in monitoring cartilage disease progression and response to therapy in arthritis. Considerable work has been published describing the methodology and validation of MRI cartilage volume measurements in healthy subjects [4, 5] and synthetic models [6], and also in models of arthritis in animals [7, 8] and osteoarthritis (OA) patient cohorts [9–12]. In RA, qualitative MRI methods have been employed to classify the stage of disease by clinically scoring the extent of synovial proliferation, together with scoring of cartilage lesions and bony erosions at the wrist and finger joints [13]. Quantitative MRI measurements of synovial joint effusion volumes and synovitis at the knee joint have also been reported [14, 15]. One report has appeared demonstrating a correlation between cartilage volume in RA patients, and Larsen grade, ESR, and CRP [16], however to our knowledge no data on reproducibility or progressive changes in cartilage volume in the knee in RA have been reported.

Volume assessment usually requires precise and accurate identification of the extent of the cartilage via the process of image segmentation. However, segmentation is technically challenging in the articular cartilage, because of the large surface area to volume ratio, the complex patterns of signal intensity variation exhibited even by normal cartilage using typical clinical MRI pulse sequences, and the additional complexity of the images in the presence of arthritis pathology. The complexity of cartilage MRI segmentation is further enhanced in RA due to the presence of pathological tissue such as joint effusion and pannus that are commonly found in the vicinity of the cartilage. Furthermore, widespread erosive damage to the cartilage is commonly observed in RA in contrast to other arthritides such as OA where damage to the cartilage is usually more focal.

The objective of this study was to quantify the various factors that influence the reproducibility of cartilage volume measurements in RA using MRI, and to explore the value of MRI in a 1-year follow-up study of RA patients.

	Method and materials

Top Abstract Introduction Method and materials Results Discussion Conclusion References

 
Patients and imaging
Following the approval of the United Bristol Healthcare Trust Ethics Committee, 23 fully informed patient volunteers were recruited for imaging after providing written consent. Inclusion criteria were a diagnosis of RA together with knee pain plus one of joint line tenderness, an obvious effusion or soft tissue swelling. A full clinical assessment of each patient was performed by a Consultant Rheumatologist (PAD), and an individual knee was selected for imaging in each case. Imaging was performed on a 1.0 T clinical scanner (‘Impact Expert’; Siemens, Erlangen, Germany) using a three-dimensional (3D) spoiled gradient-echo (FLASH) sequence with fat suppression. The imaging parameters used were: TR (sequence repetition time) 50 ms, TE (echo time) 11 ms, 40° nutation angle, and a 100 mm slice selection slab which generated 64 contiguous sagittal slices across the knee, each 1.56 mm thick. The image field-of-view in each case was 140 mm x 140 mm; images were acquired with 192 phase encode steps, and 256 frequency encode steps, and zero-filled to allow display on a 256 x 256 matrix. The displayed in-plane pixel resolution was 0.55 mm x 0.55 mm. An image time of 10 min (one signal average) was chosen, in order to minimize patient movement artefact. Previously, Morgan et al [17] established the reproducibility of this imaging protocol and found no relevant systematic differences in knee cartilage volume measurement from MR images acquired with this protocol using scanners at 1.0 T or 1.5 T.

Five substudies were performed (Table 1). The group in substudy 3 were scanned at baseline and again within an hour after cessation of knee weight-bearing without reimposition of weight-bearing. All patients were scanned during the middle of the day on each occasion.

Image analysis
After data acquisition, an image analysis package (Tosca; IBM, Winchester, UK) was used to segment all four hyaline cartilage compartments (femoral, medial tibial, lateral tibial and patellar) of one knee.

In previous studies with healthy joints [22], we employed a semi-automatic image segmentation technique in which the segmenter placed within the cartilage a seed point, which invoked a region-growing algorithm [23]. However, preliminary experience of segmentation for this patient population (some with advanced RA) using that method proved the semi-automated approach to be unreliable. In many cases the algorithm was incapable of distinguishing between genuine cartilage signal and surrounding soft tissue effusions. Consequently we elected to use a fully manual segmentation procedure. Manual segmentation was performed by two medical physicists familiar with MR images of the knee (SJG: Segmenter A and MCK: Segmenter B, details in Table 1) under the supervision of a musculoskeletal radiologist (IW). Femoral, patellar, medial tibial and lateral tibial cartilage segmentations were generated by enclosing all cartilage pixels within a region-of-interest (ROI) for each 2D image slice. Following careful checking and realignment of the ROI contour where necessary, the process was repeated for each image slice until a cartilage compartment volume had been fully segmented. The process of segmenting an entire knee (four compartment segmentations) with thorough checking took approximately 4–5 h.

We calculated the volume of each segmentation by finding the area of each 2D ROI and multiplying by the slice thickness over all slices. The Tosca package uses a curve-fitting algorithm in order to interpolate a contour defining each 2D ROI. We found the area of each enclosed 2D ROI by using this curve-fitting algorithm and then re-sampling this interpolated curve with 1000 equally-spaced vertices to produce a dense polygon. The area of a polygon with n vertices (x_i, y_i) {i=0..n–1} is then given by:

In addition to the quantitative volume calculation, 3D volume images were reconstructed and displayed using the data visualization package DX data explorer (IBM, Winchester, UK).

Effect of image quality on reproducibility
Prior to segmentation, in order to assess the effect of image quality on reproducibility, both segmenters studied the quality of all patient images at all timepoints for evidence of movement artefact and "ease of segmentation". A score for movement artefact was recorded on a qualitative 1–5 scale, where 1=very sharp, 2=sharp, 3=fair, 4=blurred and 5=very blurred. Similarly, ease of segmentation was assessed by taking into account features such as cartilage structural integrity, osteophytes, synovitis and joint effusions. Ease of segmentation was scored on a qualitative 1–5 scale, where 1=very easy, 2=easy, 3=fair, 4=difficult and 5=very difficult.

Segmentation rules
In order to provide experimental consistency, both operators agreed upon a segmentation strategy for dealing with any potential pitfalls prior to the study. The common difficulties were agreed to be erosive loss of cartilage and identification of cartilage in regions of rheumatoid effusion or synovitis. During segmentation, care was taken to identify regions of cartilage thinning and lesions, i.e. regions of complete (or very nearly complete) loss. Any regions of cartilage associated with damaged bony surfaces were carefully examined before inclusion into (or exclusion from) the segmentation. Similarly, all regions of cartilage in close proximity to rheumatoid tissue were carefully reviewed, in order to ensure that only cartilage was included in the ROI.

Geometrical issues (e.g. partial volume effects) were also carefully considered before embarking upon segmentation. To avoid errors due to partial volume effects, each operator only segmented image slices where an articulating surface could be clearly defined. To aid segmenter interpretation of partial volume effects, a 3D model of the knee joint was provided for anatomical reference (Somso NS50; Adam Rouilly, Sittingbourne UK).

Statistical analysis of the data
Errors in measuring cartilage volumes may arise from many different experimental or physiological sources. Such errors can be assessed either through a single experiment with nested analysis of variance in which all possible sources of error are systematically varied, or through a series of experiments with pairwise comparison in which each possible source of error is varied in turn. The latter approach was adopted in this work. The reproducibility of each pairwise comparison was assessed from the test–retest coefficient of variation (CoV). For each subject, s, the CoV is the standard deviation, _s, for the two measurements on that subject, divided by the mean volume, µ_s, for the subject. The overall test–retest CoV for a group of N subjects is then

Use of the CoV is most appropriate when _s is proportional to µ_s. If _s is independent of µ_s, it is appropriate to compare _s values directly. In this work it was found that _s increased with µ_s (although the dependency was less than linear), so the CoV was employed. Point estimates of CoV are imprecise unless very large numbers of patients are sampled: in the case of a small sample, it is also helpful to quote the upper bound of the 80% confidence interval on the CoV:

Changes between group means were compared using the paired two-tailed Student's t-test. A p-value of less than 0.05 was considered significant.

	Results

Top Abstract Introduction Method and materials Results Discussion Conclusion References

 
General observations
All subjects gave analysable data: no datasets were excluded from quantitative analysis. A typical MR image and the corresponding volume segmentation are shown in Figure 1. These images clearly highlight the MRI features of the disease process, and illustrate the difficulties in segmenting. Segmentation was found to be particularly difficult when dealing with the following situations:

• Regions of very thin cartilage, where the "band structure" was lost or altered;
  
• Regions of "merged" cartilage between two articulating surfaces (e.g. central regions of the tibiofemoral joint, and the base of the patellofemoral joint);
  
• Regions of cartilage above and surrounding bony abnormalities, such as erosions;
  
• Femoral cartilage at the base of the intercondylar notch, and also in other weight-bearing areas;
  
• Cartilage associated with rheumatoid synovitis, particularly in central regions of the patella and posterior regions of the femur.
  

View larger version (60K): [in this window] [in a new window]   Figure 1. (A) A typical sagittal image slice taken from a patient with rheumatoid arthritis, showing cartilage loss (arrow i), bony erosion (arrow ii) and joint effusion (arrow iii). (B) The resulting three-dimensional (3D) segmentation of the femoral cartilage, showing cartilage loss (arrow iv).

Repeat scans of the same patients in the same visit (<1 h interval, substudy 3) or in different visits (1 week interval, substudy 4) also demonstrated good reproducibility, although this was generally inferior to the same-segmenter-same-scan comparison (Table 3). In comparison with the first scan, total cartilage volume was 4% higher in substudy 3 and 0.6% lower in substudy 4 at the second scan. These tiny differences were not statistically significant. A relationship between standard deviation (SD) and mean was found for intrasegmenter, intersegmenter and scan–scan comparisons. CoV tended to decline with increasing volume with approximately two-thirds power relationship consistent with the hypothesis that the errors in this case depend strongly on the surface area characteristics of the segmentation, since the surface area tends to increase with the two-thirds power of volume.

Progression
Table 4 shows the demographic information for the 17 subjects included in substudy 5 (progression over 1 year). All subjects met the inclusion criteria. In addition to pain in the index knee all had swelling (17/17); the majority had knee effusions (12/17) and a few had a tender joint (4/17). Assessment of the X-rays showed that five of the 17 subjects had osteophytes, while six of the 17 had erosions. The mean change in joint space width over the 12 months of the study was –0.26 mm (SD 0.32, n=15).

View larger version (13K): [in this window] [in a new window]   Figure 2. Total cartilage volume in the knee in 17 rheumatoid arthritis patients over 12 months ( male; • female).

	Discussion

Top Abstract Introduction Method and materials Results Discussion Conclusion References

Variations in cartilage volume determined by MRI can be influenced by scan quality, by the choice of segmentation technique, by observer segmentation bias, and by swelling or hydration in response to short term local changes in pressure. Other variations in the measured volume may be attributable to biological change as a result of disease processes.

In this study we adopted a manual segmentation technique, and explored these potential causes of volume variation. This manual method was implemented to extract the cartilage from surrounding tissue with similar MR signal intensities, which was not possible with automated or semi-automated methods. We achieved a "best case" CoV of 2.2% (intrasegmenter measurement, total cartilage volume; Table 3) for our patient cohort, which is similar to measurements made by ourselves and others on healthy joints [22, 28, 29]. This best case CoV was unsurprising, since errors that are made assigning cartilage at the point of articulation to individual compartments are effectively eliminated when the entire knee volume is considered. In virtually all cases, the intersegmenter CoV was larger than the intrasegmenter CoV. This has important implications for clinical studies where longitudinal cartilage volume assessments may be required. This major source of measurement error may be eliminated by using the same segmenter for intrapatient comparisons.

CoVs were generally smaller for measurements made on male joints compared with female joints, which reflects the larger (and therefore simpler to segment) anatomy of the male joints. Further, the CoVs for measurements made on these RA joints were mostly larger than those reported previously on OA joints [30]. This highlights the particular difficulty in RA that can arise when attempting to distinguish cartilage from surrounding diseased tissues such as joint effusion or pannus. However, somewhat surprisingly, reproducibility did not appear to deteriorate when image quality was suboptimal or "segmentation difficulty" increased. Qualitative scoring of "movement artefact" and ease of segmentation was also undertaken for a cohort of 16 OA patients as part of a previously published parallel study [9]. The median movement artefact scores for OA and RA cohorts were identical when compared (the images were deemed sharp), whilst the median ease of segmentation score was, as expected, higher for RA (3, fair) than for OA (2, easy). This demonstrates that the quality of the acquired data is reproducible between studies, but that the ease of segmentation due to the presence of pathological tissue is variable.

Repeated measurement of cartilage volumes within the same scan session (substudy 3) and within 1 week (substudy 4) was also found to be reproducible. The CoVs resulting from these substudies are comparable with those obtained for intrasegmenter measurements made on the same datasets. In other words, short-term volume fluctuations are negligible, and our instrument and measurement techniques are stable.

Baseline cartilage volumes for these patients with RA were found to be smaller than those for normal control subjects or from a similar OA population [9], but comparable with (slightly larger than) the RA volumes reported in the recent study by Kimura [16]. In our study, no significant changes to the cartilage volumes were seen when patients were followed over 12 months. The largest change was observed in the lateral tibial compartment where a loss of 7% was recorded over 12 months. However, the CoV associated with the volume measurement for different scans and different visits (substudy 4) for this compartment was 8.7%, which implies that a 7% loss over 12 months cannot be considered significant. Furthermore, if the total cartilage volume is considered then the net loss over 12 months was only 1%. A weak correlation between cartilage volume and joint space narrowing was evident when a one-sided comparison was implemented, i.e. we assumed that a loss in cartilage volume did not correlate with an increase to the joint space width. However, Mazzuca et al [31] have reported that a reduction in joint pain can result in alteration of the positioning of the knee (in particular favouring a better knee full extension), resulting in a change in joint space width that may confound the estimate of JSN. This potential source of error was not evaluated in our study, and as such a two-sided comparison may be considered more appropriate. However we believe that the likelihood of this reciprocal correlation would be extremely low.

Clinical assessment of joint tenderness, effusions and swelling were also recorded for each patient in the study. There was no difference in median score of tenderness or effusions in those individuals who demonstrated no loss of cartilage relative to those who demonstrated a small loss. However there was a trend towards increased swelling in this latter group, but the significance of these results is unknown given the low sample size and the small cartilage volume loss. In summary, whilst we acknowledge that our work has only considered data from a relatively small patient cohort, even with limited patient numbers we are satisfied that the study methodology is robust, and that these volume measurements have remained static over the 12 month duration.

	Conclusion

Top Abstract Introduction Method and materials Results Discussion Conclusion References

 
We have considered the main variables that affect reproducibility in measuring cartilage volume changes in RA. The intersegmenter component of volume measurement is clearly the greatest source of variability, which suggests that the same segmenter should be used for any intrapatient longitudinal studies of cartilage volume. We have found that our method of volume measurement is reproducible, even when RA knees are compared over weekly durations. These session-to-session CoVs are comparable with the intrasegmenter measurement CoVs made on the same dataset. The CoV for the entire (total) cartilage is consistently smaller than for any of the individual compartments, which suggests that if cartilage loss is uniform across compartments then the "total" compartment offers the best statistical power. Over the 12 month duration of the study there was no significant change in total cartilage volume, nor were there significant changes in volume in any individual compartment.

	Acknowledgments

 
The authors would like to acknowledge Bristol University Rheumatology Unit staff for their help with patient recruitment and care, and all the MRI Radiography staff at Bristol Royal Infirmary for scanning the patients.

	Footnotes

 
Current address for Dr Stephen J Gandy, Medical Physics Department, Ninewells Hospital, Dundee DD1 9SY, UK.

Current address for Dr A D Brett, Image Metrics plc, Regent House, Heaton Lane, Stockport SK4 1BS, UK.

Current address for Dr I Watt, Department of Radiology C2-S, Leids Universitair Medisch Centrum, Postbus 9600, 2300 RC Leiden, The Netherlands.

Grant sponsors: Medical Research Council, AstraZeneca, United Bristol Healthcare NHS Trust.

Received for publication May 10, 2004. Revision received August 12, 2004. Accepted for publication September 7, 2004.

	References

Top Abstract Introduction Method and materials Results Discussion Conclusion References

 

1. Larsen A, Dale K, Eek M. Radiographic evaluation of rheumatoid arthritis and related conditions by standard reference films. Acta Radiol Diagn 1977;18:481–91.
2. Sharp JT, Young DY, Bluhm GB, Brook A, Brower AC, Corbett M, et al. How many joints in the hands and wrists should be included in a score of radiologic abnormalities used to assess rheumatoid arthritis? Arthritis Rheum 1985;28:1326–35.[Medline]
3. Van der Heijde DM. Radiographic imaging: the ‘gold standard’ for assessment of disease progression in rheumatoid arthritis. Rheumatology (Oxford) 2000;39 Suppl. 1:9–16.
4. Cicuttini FM, Wluka AE, Wang Y, Davis SR, Hankin J, Ebeling P. Compartment differences in knee cartilage volume in healthy adults. J Rheumatol 2002;29:554–6.[Abstract/Free Full Text]
5. Stammberger T, Eckstein F, Michaelis M, Englmeier K-H, Reiser M. Inter-observer reproducibility of quantitative cartilage measurements: comparison of B-spline snakes and manual segmentation. Magn Reson Imaging 1999;17:1033–42.[CrossRef][Medline]
6. Kauffmann C, Gravel P, Godbout B, Gravel A, Beaudoin G, Raynauld JP, et al. Computer-aided method for quantification of cartilage thickness and volume changes using MRI: validation study using a synthetic model. IEEE Trans Biomed Eng 2003;50:978–88.[CrossRef][Medline]
7. Tessier JJ, Bowyer J, Brownrigg NJ, Peers S, Westwood FR, Waterton JC, et al. Characterisation of the guinea pig model of osteoarthritis by in-vivo three-dimensional magnetic resonance imaging. Osteoarthritis Cartilage 2003;11:845–53.[CrossRef][Medline]
8. Faure P, Doan BT, Beloeil JC. In-vivo high resolution three-dimensional MRI studies of rat joints at 7 T. NMR Biomed 2003;16:484–93.[CrossRef][Medline]
9. Gandy SJ, Dieppe PA, Keen MC, Maciewicz RA, Watt I, Waterton JC. No loss of cartilage volume over three years in patients with knee osteoarthritis as assessed by magnetic resonance imaging. Osteoarthritis Cartilage 2002;10:929–37.[CrossRef][Medline]
10. Vanwanseele B, Eckstein F, Knecht H, Spaepen A, Stussis E. Longitudinal analysis of cartilage atrophy in the knees of patients with spinal cord injury. Arthritis Rheum 2003;48:3377–81.[CrossRef][Medline]
11. Cicuttini FM, Wluka AE, Wang Y, Stuckey SL. Longitudinal study of changes in tibial and femoral cartilage in knee osteoarthritis. Arthritis Rheum 2004;50:94–7.[CrossRef][Medline]
12. Peterfy C, White D, Zhao J, VanDijke C, Genant HK. Longitudinal measurement of knee cartilage volume in osteoarthritis (Abstract). Arthritis Rheum 1998;41:S361.[CrossRef]
13. McQueen F, Lassere M, Edmonds J, Conaghan P, Peterfy C, Bird P, et al. OMERACT Rheumatoid arthritis magnetic resonance imaging studies. Summary of OMERACT 6 MR imaging module. J Rheumatol 2003;30:1387–92.[Abstract/Free Full Text]
14. Gylys-Morin VM, Brent Graham T, Blebea JS, Dardzinski BJ, Laor T, Johnson ND, et al. Knee in early juvenile rheumatoid arthritis: MR imaging findings. Radiology 2001;220:696–706.[Abstract/Free Full Text]
15. Ostergaard M, Ejbjerg B, Stoltenberg M, Gideon P, Volck B, Skov K, et al. Quantitative magnetic resonance imaging as a marker of synovial membrane regeneration and recurrence of synovitis after arthroscopic knee joint synovectomy: a one year follow up study. Ann Rheum Dis 2001;60:233–6.[Abstract/Free Full Text]
16. Kimura M. Quantification of cartilage volume in the knees of rheumatoid arthritis patients: relationship of physical characteristics and disease conditions to cartilage volume. Teikyo Med J 2003;26:411–23.
17. Morgan SR, Waterton JC, Maciewicz RA, Leadbetter JE, Gandy SJ, Moots RJ, et al. Magnetic resonance imaging measurement of knee cartilage volume in a multicentre study. Rheumatology 2004;43:19–21.[Abstract/Free Full Text]
18. Dieppe PA, Cushnaghan J, Young P, Kirwan J. Prediction of the progression of joint space narrowing in osteoarthritis of the knee by bone scintigraphy. Ann Rheum Dis 1993;52:557–63.[Abstract/Free Full Text]
19. Dieppe PA, Cushanghan J, Shepstone L. The Bristol ‘OA 500’ study: progression of osteoarthritis (OA) over 3 years and the relationship between clinical and radiographic changes at the knee joint. Osteoarthritis Cartilage 1997;5:85–97.
20. Mazzuca SA, Brandt KD, Dieppe PA, Doherty M, Katz BP, Lane KA. Effect of alignment of the medial tibial plateau and x-ray beam on apparent progression of osteoarthritis in the standing anterioposterior knee radiograph. Arthritis Rheum 2001;44:1786–94.[CrossRef][Medline]
21. Lequesne M. Quantitative measurements of joint space during progression of osteoarthritis: chondrometry. In: Kuettner K, Goldberg V, editors. Osteoarthritic disorders. Rosemont, IL: American Academy of Orthopaedic Surgeons, 1995:427–44.
22. Waterton JC, Solloway S, Foster JE, Keen MC, Gandy S, Middleton BJ, et al. Diurnal variation in the articular cartilage of the knee in young adult humans. Magn Reson Med 2000;43:126–32.[CrossRef][Medline]
23. Sivewright GJ, Elliott PJ. Interactive region and volume growing for segmenting volumes in MR and CT images. Medical Inf (Lond) 1994;19:71–80.
24. Kirwan J, Byron M, Watt I. The relationship between soft tissue swelling, joint space narrowing and erosive damage in hand x-rays of patients with rheumatoid arthritis. Rheumatology 2001;40:297–301.[Abstract/Free Full Text]
25. Bird P, Lassere M, Shnier R, Edmonds J. Computerized measurement of magnetic resonance imaging erosion volumes in patients with rheumatoid arthritis: a comparison with existing magnetic resonance imaging scoring systems and standard clinical outcome measures. Arthritis Rheum 2003;48:614–24.[CrossRef][Medline]
26. Ostergaard M, Hansen M, Stoltenberg M, Jensen KE, Szkudlarek M, Pedersen-Zbinden B, et al. New radiographic bone erosions in the wrists of patients with rheumatoid arthritis are detectable with magnetic resonance imaging a median of two years earlier. Arthritis Rheum 2003;48:2128–31.[CrossRef][Medline]
27. McQueen FM, Benton N, Perry D, Crabbe J, Robinson E, Yeoman S, et al. Bone edema scored on magnetic resonance imaging scans of the dominant carpus at presentation predicts radiographic joint damage of the hands and feet six years later in patients with rheumatoid arthritis. Arthritis Rheum 2003;48:1814–27.[CrossRef][Medline]
28. Peterfy CG, van Dijke CF, Janzen DL, Glüer CC, Namba R, Majumdar S, et al. Quantification of articular cartilage in the knee with pulsed saturation transfer subtraction and fat-suppressed MR imaging: optimization and validation. Radiology 1994;192:485–91.[Abstract/Free Full Text]
29. Westhoff J, Eckstein F, Sittek H, Lösch A, Faber S, Kolem H, et al. Dreidmensionale dicken- und volumenbestimmung des knie-gelenkknorpels mit der MRT: reproduzierbarkeit am probanden. Fortschr Röntgenstr 1997;167:585–90.[CrossRef]
30. Eckstein F, Cicuttini F, Raynauld J-P, Waterton JC, Peterfy C. Magnetic resonance imaging (MRI) of articular cartilage in knee osteoarthritis (OA): morphological assessment. Osteoarthritis Cartilage 2004; (In press).
31. Mazzuca SA, Brandt KD, Lane KA, Katz BP. Knee pain reduces joint space width in conventional standing anteroposterior radiographs of osteoarthritic knees. Arthritis Rheum 2002;46:1223–7.[CrossRef][Medline]

This article has been cited by other articles:

				F P P Martins, B Gutfilen, S A L de Souza, M N L de Azevedo, L R Cardoso, R Fraga, and L M B da Fonseca Monitoring rheumatoid arthritis synovitis with 99mTc-anti-CD3 Br. J. Radiol., January 1, 2008; 81(961): 25 - 29. [Abstract] [Full Text] [PDF]

				R. J. Hodgson, P. O'Connor, and R. Moots MRI of rheumatoid arthritis image quantitation for the assessment of disease activity, progression and response to therapy Rheumatology, January 1, 2008; 47(1): 13 - 21. [Abstract] [Full Text] [PDF]

				C. Schirmer, A. K Scheel, C. E Althoff, T. Schink, I. Eshed, A. Lembcke, G.-R. Burmester, M. Backhaus, B. Hamm, and K.-G. A Hermann Diagnostic quality and scoring of synovitis, tenosynovitis and erosions in low-field MRI of patients with rheumatoid arthritis: a comparison with conventional MRI Ann Rheum Dis, April 1, 2007; 66(4): 522 - 529. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Gandy, S J Articles by Watt, I Search for Related Content PubMed PubMed Citation Articles by Gandy, S J Articles by Watt, I