Assessment of Femoral Cartilage Morphological and Topological Features Using Machine Learning

Cartilage is an important interface between bones in the human body, but is often prone to deterioration, especially within an aging population [1]. Understanding the topology and morphology mathematically can aid and ease the diagnosis of degenerative diseases, relieving stress on the health care systems by reducing time taken to diagnose. This research explores the topological and morphological features of femoral cartilage using computer-aided design (CAD) models and machine learning (ML) methods to classify degeneration experienced in patient cartilage. Features extracted are related from a three dimensional wall-thickness analysis, two dimensional thickness measurements, as well as fat and water content within the cartilage. The preliminary analysis result in abnormally accurate models for classifying degeneration, but should be taken with a grain of salt as the data-set only includes 46 patients. The accuracies range from 83 to 91%, therefore the next steps would be to expand the data-set, re-train the models and analyze the results to see the effects and the classification accuracy.