Objective Assessment of Joint Damage

Abstract

Determining a composite score includes deriving, based on an image set that includes at least one three-dimensional image of the synovial joint, first and second information. The first information indicates cumulative damage to the synovial joint. The second information indicates either one or both joint pain and loss of function of the synovial joint. The resulting composite score provides an objective measure of joint damage or an extent of joint disease.

Claims

1. A method comprising determining a composite score that provides an objective evaluation of overall severity of a disease of a synovial joint, wherein determining said composite score comprises receiving an image set, said image set including at least one image of said synovial joint; based on said image set, deriving first information and second information, said first information being indicative of cumulative damage to said synovial joint and said second information being indicative of at least one of joint pain and loss of function of said synovial joint; and based at least on said first and second information, determining said composite score, said composite score being an objective measure of overall severity of the synovial joint disease.

2. (canceled)

3. The method of claim 1, wherein said first information is indicative of cartilage damage within said synovial joint.

4. The method of claim 1, wherein said second information is indicative of a volume of said synovial joint that has been affected by a feature indicative of joint damage.

5-13. (canceled)

14. The method of claim 1, further comprising determining progression of joint damage over a period of time, wherein determining said progression comprises repeating said determination of said composite score multiple times during said period.

15-16. (canceled)

17. The method of claim 1, wherein said first information is indicative of cartilage damage and wherein said method further comprises evaluating a summation of weighted measurements, each of said measurements being obtained at pre-specified informative locations on said synovial joint, wherein weighted measurements are weighted based at least in part on a size of said synovial joint.

18. The method of claim 1, wherein said first information is indicative of cumulative damage and said second information is indicative of disease activity.

19-25. (canceled)

26. The method of claim 1, further comprising enriching a clinical trial based at least in part on said composite score.

27. The method of claim 1, further comprising determining, based at least in part on said composite score, that an overall status of said joint has changed.

28. The method of claim 1, further comprising determining, based at least in part on said composite score, that there has been a change in an extent of activity of said disease.

29. The method of claim 1, further comprising determining, based at least in part on said composite score, that there has been a change in an extent of activity said cumulative damage

30. The method of claim 1, further comprising, based at least in part on said composite score, predicting an extent to which structural changes in said joint will progress.

31. The method of claim 1, further comprising, based at least in part on said composite score, predicting an extent to which pain resulting from said disease will progress.

32. The method of claim 1, further comprising, based at least in part on said composite score, predicting an extent to which functional impairment resulting from said disease will progress.

33. The method of claim 1, further comprising, based at least in part on said composite score, predicting a likelihood that said disease will result in a need for replacement of said joint.

34. The method of claim 1, further comprising, based at least in part on said composite score, predicting whether said disease is progressing.

35. The method of claim 1, further comprising carrying out an intervention to ameliorate said disease and using said composite score as a biomarker upon which to base a determination that said intervention has changed the course of said disease.

36. The method of claim 1, wherein determining said composite scores comprise weighting either the first information or the second information.

37-45. (canceled)

46. A method comprising acquiring a first set of images, each of which corresponds to a location of cartilage within a synovial joint; using said first set of images, determining a cartilage-damage index for each of said locations; determining a cumulative-damage score based on said cartilage-damage index for each of said locations, acquiring a second set of images, said images in said second set being images of volumes having bone-marrow lesions and images of volumes affected by effusion from synovitis volumes within said synovial joint; using said second set of images, determining a disease-activity score; and determining a composite score based on said cumulative damage score and said disease-activity score.

47. The method of claim 46, further comprising, based at least in part on said composite score, predicting progression of a disease of said joint.

48. (canceled)

49. An apparatus comprising a joint evaluator that receives images of a synovial joint of a subject, wherein said joint evaluator comprises a volumetric analyzer and a deterioration analyzer, wherein said volumetric analyzer is configured to extract, from said images, an estimate of volume occupied by certain features that are indicative of said synovial joint's condition, and wherein said deterioration analyzer is configured to extract, from said images information indicative of physical deterioration of said synovial joint.

Description

DESCRIPTION OF DRAWINGS

[0053] FIG. 1 shows a joint evaluator coupled to an MRI machine and

[0054] FIG. 2 shows the components of the joint evaluator of FIG. 1.

DESCRIPTION

[0055] Referring to FIG. 1, a joint evaluator 10 receives images 12 of a joint of a subject 16 who has been transferred into a magnetic-resonance imaging machine 18 from a patient table 21. The magnetic-resonance imaging machine 18 includes a magnet 23, gradient coils 25, an RF coil 27, and a scanner 29.

[0056] As shown in FIG. 2, the joint evaluator 10 includes a volumetric analyzer 20 and a deterioration analyzer 22 that each receive the images 12.

[0057] The volumetric analyzer 20 extracts, from the images 12, an estimate of the volume occupied by certain features that are indicative of the joint's condition. Examples of suitable volumes include the volume occupied by bone-marrow lesions and the volume affected by effusion resulting from synovitis. A suitable volumetric analyzer 20 is one that uses signal intensity thresholds to delineate various structures of interest and to estimate or measure the volumes of those structures.

[0058] The deterioration analyzer 22 extracts, from the images 12, information indicative of physical deterioration of the joint itself. In the case of the knee, the deterioration analyzer 22 outputs an index indicative of cartilage damage in the patella and/or damage to the medial and lateral tibiofemoral cartilage.

[0059] A composite-score generator 24 receives one or more deterioration scores 28 from the deterioration analyzer 22 and one or more volumes 26 from the volumetric analyzer 20 and uses them to derive a suitable composite knee score 30.

[0060] The operation of the joint evaluator 10 can be understood with reference to a particular application in which the joint evaluator 10 was applied to develop quantitative composite-knee scores using baseline data and data obtained two years later, which was selected from a database maintained by the Osteoarthritis Initiative. The Osteoarthritis Initiative is a multicenter cohort study of subjects 16 who are within the United States and who either already have symptomatic knee osteoarthritis or are at risk of developing it.

[0061] The joint evaluator 10 was used to inspect joints of subjects 16 who had had magnetic-resonance readings of their joint assessed as part of a nested case-control study to evaluate longitudinal change and who had already provided baseline data and two-years of longitudinal data for clinical, radiographic, and magnetic-resonance outcomes. A development dataset and a validation dataset were selected via stratified random sampling based on a baseline Kellgren-Lawrence grade and progression within baseline Kellgren-Lawrence strata.

[0062] The development dataset was used to evaluate candidate quantitative composite knee-scores 30 and to select a subset of the best-performing composite knee-scores 30 to move onto further testing in the validation dataset.

[0063] The validation dataset was used to further validate the quantitative composite knee-scores 30 and select those that had the best discriminative and predictive ability.

[0064] In the foregoing study, the deterioration analyzer 22 carried out semi-automated measurements to obtain an index of articular cartilage damage. The volumetric analyzer 20 was used in connection with semi-automated measurements of volumes of bone-marrow lesions and volumes affected by effusion resulting from synovitis. Both the deterioration analyzer 22 and the volumetric analyzer 20 relied on images 12 obtained by a magnetic-resonance imaging machine 18.

[0065] The deterioration analyzer 22 derived the cartilage-damage index from measurements made by three-dimensional dual echo steady state imaging with a 140-millimeter field-of-view, 0.7-millimeter thick slices, a zero-millimeter skip, a 25-degree flip angle, an echo time of 4.7 milliseconds, a recovery time of 16.3 milliseconds, and a 307×384 matrix with resolution of 0.365 millimeters along a first direction and a resolution of 0.456 millimeters along a second direction that was perpendicular to the first direction. The total number of slices was 160.

[0066] The volumetric analyzer 20 measured bone-marrow lesion and effusion-synovitis volumes by using the intermediate-weighted fat-suppressed imaging sequence with a slightly larger field-of-view and a smaller number of thicker slices. In particular, the volumetric analyzer 20 relied on a field-of-view of 160 millimeters, and three-millimeter thick slices. The volumetric analyzer 20 used images 12 obtained with a zero-millimeter skip, a 180-degree flip angle, an echo time of 30 milliseconds, a recovery time of 3200 milliseconds, and a 313×448 matrix with a resolution of 0.357 millimeters along the first direction and a resolution of 0.511 millimeters along the second direction. The total number of slices was thirty-seven.

[0067] The foregoing parameters are only exemplary. It would also have been possible to carry out the method using different parameters.

[0068] The deterioration analyzer 22 derived a damage index for the medial and lateral tibiofemoral cartilage. This was carried out by having the deterioration analyzer 22 inspect baseline and 2-year three-dimensional dual-echo steady-state magnetic resonance images 12 using previously validated methods. In doing so, the deterioration analyzer 22 participated in a procedure that included three steps.

[0069] In a first step, the deterioration analyzer 22 determined the medial-lateral width of the femur by selecting the medial-most and lateral-most magnetic resonance images 12 that showed bone. This first step relied on a human assistant 32 to mark the most medial and most lateral image that showed bone before the deterioration analyzer 22 could carry out its task.

[0070] These images 12 represented minimum and maximum values of the medial-to-lateral axis of the coordinate system, which will be referred to herein as the “y-axis.” The deterioration analyzer 22 automatically indicated the slices that contained informative locations in each of the medial and lateral tibiofemoral compartments based on the coordinate system.

[0071] A second step also required the human assistant 32. In this step, the assistant 32 manually traced the bone-cartilage boundary using predefined segmentation rules on each of the slices. However, this procedure could also have been carried out semi-automatically.

[0072] Following this second step, control was returned to the deterioration analyzer 22 to carry out a third step. In this third step, the deterioration analyzer 22 translated the length of the bone-cartilage boundary to a standardized anterior-to-posterior axis and also identified the informative locations so that the assistant 32 could manually carry out measurements of cartilage thickness at those locations. However, this procedure could also have been carried out semi-automatically.

[0073] Once the assistant 32 carried out this procedure, the deterioration analyzer 22 then computed the cartilage damage index. It did so by summing the products of cartilage thickness, cartilage length measured in the anterior to posterior direction, and voxel size from each informative location.

[0074] The deterioration analyzer 22 also provided an index that was indicative of damage to the cartilage of the patella. The deterioration analyzer 22 carried this out based on both baseline images and 2-year three-dimensional dual-echo steady-state magnetic resonance images.

[0075] The deterioration analyzer 22 determined the medial-lateral width of the patella by receiving, from the human assistant 32, a selection of the medial-most and lateral-most magnetic resonance image slices that included bone. Upon having received the human assistant's selection, the deterioration analyzer 22 then identified the six magnetic resonance images with informative locations, three from each of the medial and lateral facets.

[0076] The assistant 32 delineated the bone-cartilage boundary using predefined segmentation rules on each of these slices. However, this procedure could also have been carried out semi-automatically or automatically. Upon completion, control returned to the deterioration analyzer 22, which then translated the length of the bone-cartilage boundary to a standardized superior-to-inferior axis and indicated the predefined informative locations.

[0077] The assistant 32 then measured the cartilage thickness at those points and provided this information to the deterioration analyzer 22. However, this procedure could also have been carried out semi-automatically or automatically. The deterioration analyzer 22 then computed the index of damage to the patella's cartilage. It did so by summing the products of cartilage thickness, cartilage length in the anterior to posterior direction, and voxel size from each informative location.

[0078] The volumetric analyzer 20 was then used to measure the extent of volumes with bone-marrow lesions. These volumes correspond to regions of high-signal intensity within bone on intermediate-weighted fat-saturated magnetic resonance images in the medial and lateral tibiofemoral and patellar compartments.

[0079] The volumetric analyzer 20 included both a marker 34 and a filter 36. The marker 34 was configured to mark a region-of-interest. In some cases, the region is that around a bone-marrow lesion. In other cases, the region is that associated with a particular slice. The filter 36 was configured to classify the marked region by comparing a predefined threshold with a score derived from an intensity histogram distribution obtained from within the region-of-interest marked by the marker 34.

[0080] A follow-up slice used the same threshold as was used in the corresponding baseline slice. Through intervention by the assistant 32, and with the help of dual screens to simultaneously display baseline and follow-up images 12, it was possible to manually adjust the threshold and to remove those regions that did not have bone-marrow lesions.

[0081] To promote the assistant's ability to co-locate the corresponding bone-marrow lesions on baseline and follow-up images, it is useful to provide dual screens to simultaneously display baseline and follow-up magnetic-resonance images.

[0082] The volumetric analyzer 20 also measured knee effusion-synovitis volume. It did so by having the marker 34 mark the first and last slice of the knee on intermediate-weighted fat-saturated magnetic resonance images as well as the base of the patella (i.e., the proximal border), and the distal attachment of the patellar ligament on a central slice for which the patellar ligament was clearly visible.

[0083] A segmentation unit 38 automatically segmented effusion-synovitis based on an existing threshold.

[0084] The assistant 32 then performed a quality check of the segmentation unit's work. To do so, the assistant 32 manually adjusted the threshold to change the effusion-synovitis boundaries. Then the assistant 32 reviewed the automatically-generated segmentation results provided by the segmentation unit 38 and removed, from those results, any areas of high signal intensity that clearly did not arise as a result of effusion-synovitis. These included high signal intensity arising from such phenomena as subchondral cysts and blood vessels.

[0085] The composite knee-score 30 is a “composite” score because it is derived from multiple factors. Each factor by itself would generally be regarded as insufficient to indicate joint condition by itself. However, when taken together, an unexpected synergy arises with the result that the whole becomes greater than the sum of the parts. By combining these factors in the correct way, the composite knee-score 30 as described herein provide objective and reliable measures of joint condition.

[0086] The composite-score generator 24 generated the composite knee-scores 30 based on measurements of articular cartilage damage and on values affected by certain conditions and also on volumes affected by certain conditions.

[0087] The measurements of articular cartilage damage comprised measurements in the medial and lateral compartments in the femoral, tibial, and patellar regions.

[0088] The measurements of affected volumes included measurements of volumes affected by bone-marrow lesions and measurements of volume affected by effusion resulting from synovitis. The measurements of volumes affected by bone-marrow lesions were made in the medial and lateral compartments in the femur, the tibia, and the patella. The measurement of volume affected by effusion due to synovitis was a single whole-knee measurement.

[0089] All magnetic resonance measurements were corrected for knee size based on bone width. Measurements were standardized thereby placing all measurements on the same scale. The difference between the standardized baseline and follow-up measures were used to develop the composite knee-score 30. Cartilage-damage index difference scores were multiplied by −1 to be consistent with the other measures, in which higher values indicate worse disease. In some practices, standardization is carried out by subtracting the baseline mean and dividing by the baselines standard deviation. However, other standardization methods can be used, particularly when the amount of available data is sufficiently large.

[0090] A priori, three approaches were identified for developing the quantitative composite-knee score 30: principal components analysis, application of inverse variance weighting to the sum of magnetic resonance measures on the original, unstandardized measures, and calculating an unweighted sum of the standardized change of the thirteen measures. The analyses led to two novel outcome subdomains: an unweighted cumulative damage score calculated by summing the standardized change for each of the six cartilage-damage measurements and an unweighted disease activity score calculated by summing the standardized change in effusion-synovitis volume and the bone-marrow-lesion volumes for the six regions.

[0091] The method described herein included intervention by a human assistant 32 to carry out certain steps. In principle, one or more of these steps could also be carried out by a suitably programmed processor.