A Comparative Study on Diabetic Foot Ulcer Diagnosis and Classification Using Deep Learning Techniques

Diabetic foot ulcers (DFUs) are a grave ramification of diabetes, resulting in considerable morbidity, increased risk of amputation, and a substantial worldwide healthcare burden. Precise and prompt diagnosis is crucial for effective care, although it poses challenges because to the intricate and diverse visual manifestations of DFUs. This work offers a thorough comparative analysis of deep learning (DL) strategies for the diagnosis and categorisation of DLs using medical photographs. Diverse schemes, like convolutional neural networks (CNNs), transformer-based models, and hybrid methodologies, are assessed utilising both public and proprietary datasets. Each approach is evaluated utilising common performance criteria, including accuracy, sensitivity, specificity, and F1-score. The research illustrates that sophisticated DL models attain diagnostic efficacy comparable to that of seasoned doctors, especially in assessing wound severity, infection, and ischaemia. Challenges persist regarding dataset quality, model generalisability, and clinical integration; yet, DL demonstrates significant potential in automating and enhancing DL diagnostic procedures. The results offer pragmatic advice for academics and clinicians in choosing successful DL algorithms to enhance diabetic wound care and patient outcomes.