Magnetic Resonance Imaging-Based Estimation of Knee Cartilage Thickness with MATLAB

Detection of early knee osteoarthritis remains a driving force in the search for more promising quantitative assessment approaches. Apart from other conventional methods such as radiography, computed tomography, and sonography, magnetic resonance imaging has become more widely available and has made it essential to visualize the knee's entire anatomy. Biomarkers such as joint space narrowing, articular cartilage thickness, cartilage volume, cartilage surface curvature, lesion depth, and others are used to determine disease progression in non-invasive manner. In this research, a regional cartilage normal thickness approximation (RCN-ta) model was developed with MATLAB to enable rapid cartilage thickness assessment with a simple click. The model formulated was compared to the FDA-cleared software measurements. A reasonable range of 0.135-0.214 mm of root-mean-square error may be predicted from the model. With a high ICC > 0.975, the model was highly accurate and reproducible. A good agreement between the proposed model and the medically used software can be found with a high Pearson correlation of r > 0.90.