Multivariate Feature-based Analysis of the Diabetic Foot Ulcers Using Machine Learning Classifiers

Diabetic foot ulcers (DFUs) are a serious consequence of diabetes mellitus, sometimes resulting in infection and amputation. Accurate and early classification of DFUs is critical for effective treatment and management. This study presents a multivariate feature-based approach to classify diabetic foot ulcers using machine learning based algorithms. The proposed methodology involves the extraction of comprehensive features from DFU images, encompassing texture, color, shape, and others. Several machine learning algorithms, including K-Nearest Neighbour (KNN), Support Vector Machines (SVM) and Back Propagation Neural Network (BPNN) were evaluated for their effectiveness in classifying DFUs into categories such as Normal and Abnormal. The dataset used in this study comprised of annotated DFU images and corresponding clinical data from a diverse patient cohort. The experimentation results of the K-Nearest Neighbour, Support Vector Machine, and Backpropagation Neural Network algorithms were 92.81%, 96.07%, and 99.04% respectively. Results indicated that BPNN outperformed other classification models, achieving an accuracy of 99.04%, which is significantly higher than other models. The integration of multivariate features significantly enhanced the model’s ability to distinguish between different DFU stages compared to using single-modal features alone. This research highlights the capability of advanced machine learning techniques in improving the diagnostic process for DFUs, ultimately contributing to better patient outcomes and optimized treatment strategies.