Advancing Diabetic Foot Ulcer Detection: A Study of Thermographic Imaging with Faster Vision Transformer

Background: Diabetic foot ulceration (DFU), a serious diabetes mellitus complication, is associated with high risks of infection, limb amputation, and life-threatening events. Objective: The present study explores the efficacy of deep learning in enhancing DFU diagnosis using infrared thermography imaging, with a focus on overcoming challenges in comprehensively evaluating plantar temperature differences—a determining factor among diabetic patients. Methods: By comparing deep learning architecture with traditional machine learning methods, the study aims to identify the most effective approaches for early detection and intervention to prevent complications. The Faster Vision Transformer (FasterViT) is employed as the basic architecture due to its established efficiency in processing medical imaging data and its ability to rapidly extract clinically relevant features, in line with the timeliness required in real-world clinical practice. Findings: FasterViT achieved a breathtaking classification accuracy of 98%, a pointer to its potential in revolutionizing DFU diagnosis. Novelty and applications: The findings underscore the revolutionary role of AI-based systems in improving diagnostic accuracy, enabling timely clinical decision-making, and, by extension, advancing patient care outcomes in diabetes management.