Grading of Femoral and Tibial Cartilage Injuries in Knee Joint Based on MRI Radiomics Features

Motivation: To create a tool that helps doctors quickly and accurately assess the severity of knee cartilage damage based on MRI images. Goal(s): Develop a nomogram for predicting the severity of cartilage injuries in knee osteoarthritis patients. Approach: Utilize deep learning to segment knee joint cartilage from MRI and extract radiomics features, combined with clinical variables to predict cartilage grading. Results: The nomogram models for femoral and tibial cartilage grading showed high accuracy, with AUC values of 0.872 and 0.914 respectively, and were validated internally. Impact: The constructed models provide a reliable method for diagnosing cartilage damage severity, assisting in designing personalized treatment plans for patients with knee osteoarthritis.