Background Osteoarthritis (OA) is the most common form of arthritis and affects disproportionately the knee. Recent developments in imaging techniques showed that OA is not just a joint disease but also involves progressive changes in the subchondral/subarticular bone area of the tibia. On top of the accepted method of measuring the joint space width, assessments of the trabecular bone structure in selected regions of interest (ROI) in conventional X-rays may be offering an alternative method for quantifying the risk and progression of this disease.
Objectives The accepted method for assessing OA - Joint space width (JSW) and Joint Space Area (JSA) measurements - have limited capabilities in regard to early identification and reproducible follow-ups of the disease. The objective of this abstract is to evaluate the trabecular bone structure as an area for early identification of OA risk, applying texture anisotropy algorithms and subsequently comparing the results to standard JSW and JSA measurements.
Methods This study was performed using data from the Osteoarthritis Initiative. The image data set was restricted to female, Caucasian, right knee exams recorded with the same modality. Furthermore we selected exams which had a KL grade of 0 at the baseline exam and a deteriorating KL grade ≥2 at the 96 month follow up. 22 cases fulfilled these criteria and we selected 22 matching controls with no signs of OA at the 96 month follow up. The selected region of interest (ROI) for the analysis of the radiographic texture consisted of four ROIs in the subchondral tibia and one additional ROI in each femur condyle – in total 6 ROIs. For each individual ROI, the degree of texture anisotropy was calculated and compared between case/control. In addition, JSW & JSA were calculated in both groups using a proprietary software-based method (ImageBiopsy Lab, Vienna, Austria).
Results Whereas the JSW and the JSA measurements did not yield any significant differences with respect to their mean values (Cohen's d =0.139 and 0.028), the calculated texture parameters showed that differences in values between cases and controls can be found in two of the subchondral ROIs (ROI1&2) with Cohens'd values of 0.625 and 0.831, respectively. With respect to selected patient, the differences in anisotropy results were significant using these texture parameters.
Conclusions Our results indicate that using the selected radiographic texture parameters, an early identification of patients at risk for developing OA using conventional X-rays can be achieved. This may offer an additional method for quantifying the risk of baseline OA. This is supported by the Cohen's d values that are by definition relatively large (0.625 and 0.831). Ongoing research focuses on larger sample set validation and the use of such algorithms for additional applications, such as the early identification of fracture risk.
Disclosure of Interest None declared