Enhanced Diabetic Foot Ulcer Detection with Deep Learning-Based Predictive Systems

Diabetes mellitus is a chronic metabolic disorder affecting millions of people globally, often leading to severe complications such as diabetic foot ulcers. DFUs can progress to infections, including cellulitis and gangrene, and may ultimately result in lower-limb amputations. Distal symmetric polyneuropathy (DSP) plays a key role in the development of these ulcers by impairing sensation and promoting unnoticed injuries. This paper presents a deep learning approach using convolutional neural networks to detect DFUs at an early stage through thermal imaging. The ResNet50, EfficientNetB0, MobileNet, and DenseNet121 models were trained to classify healthy and ulcerated foot images with high accuracy. Image preprocessing techniques, including edge detection and segmentation, were applied to improve feature extraction and enhance prediction performance. The results demonstrate that these models can effectively identify the early signs of DFUs, enabling timely medical intervention. This study highlights the potential of AIdriven diagnostic systems to reduce DFU-related complications and improve patient outcomes.