Self-Detection of Diabetic Foot Ulcer Using Thermography and Deep Learning

Diabetic foot ulcers (DFUs) are a severe complication of diabetes, necessitating early detection to mitigate morbidity and mortality. Current diagnostic methods are often manual, time-consuming, and lack real-time automated solutions. This study proposes a deep learning framework leveraging thermal imaging for real-time DFU detection. Utilizing a dataset of 1,800 high-resolution thermal images ($1,440 \times 1,080$ pixels) categorized into “Healthy” and “Ulcer” classes, the framework integrates preprocessing techniques—noise reduction, contrast adjustment, and resizing—to optimize input quality. The model's performance is benchmarked against Inception V3 and MobileNet V2, evaluating accuracy, precision, recall, and computational efficiency. Results demonstrate that the proposed model achieves 100% validation accuracy and near-zero loss (0.0006%) by epoch 10, significantly outperforming Inception V3 (86.3% accuracy) and MobileNet V2 (85.55% accuracy). The confusion matrices reveal perfect classification for the proposed model, with no misclassifications, underscoring its clinical reliability.