Exploring deep learning capabilities in knee osteoarthritis case study for classification

Abstract

This research study is devoted to the investigation of deep neural networks (DNN) for classification of the complex problem of knee osteoarthritis diagnosis. Osteoarthritis (OA) is the most common chronic condition of the joints revealing a variation in symptoms' intensity, frequency and pattern. A large number of features/factors need to be assessed for knee OA, mainly related with medical risks factors including advanced age, gender, hormonal status, body weight or size, family history of disease etc. The main goal of this research study is to implement deep neural networks as a new efficient machine learning approach for this classification task taking into account the large number of medical factors affecting OA. The potential of the proposed methodology was demonstrated by classifying different subgroups of control participants from self-reported clinical data and providing a category of knee OA diagnosis. The investigated subgroups were defined by gender, age and obesity. Furthermore, to validate the proposed deep learning methodology, a comparison analysis between the proposed DNN and some benchmark machine learning techniques recommended for classification was conducted and the results showed the effectiveness of deep learning in the diagnosis of knee OA.