Radon Feature Based Osteoarthritis Severity Assessment

Knee Osteoarthritis (KOA) is a common joint degeneration identified by joint stiffness, pain, and functional disability. Physical symptoms, medical histories, and other joint screening techniques like radiography, MRIs, and CT scans are commonly taken into account while evaluating it. The traditional approaches, however, are quite subjective, which makes it difficult to detect early sickness progression. We propose, a machine-learning approach to automatically classify the severity of KOA using MRI images. Mask RCNN segments the knee’s upper and lower joints. The cartilage area is then the Region of Interest (ROI), which is acquired via morphological techniques. The radon transform is used to extract the dominating features from ROI, and the K Nearest Neighbor (KNN) classifier is used to categorize them. Comparing the experimental findings of the suggested technique to those of other machine learning classifiers and state-of-the-art methods, the proposed method outperformed them all with a classification accuracy of 88%. The results of the studies show that the suggested method aids surgeons in early diagnosis and minimizes problems associated with KOA.