ResNet-50 Transfer Learning Model for Diabetic Foot Ulcer Detection Using Thermal Images
Aditya Kumar, Leema Nelson, Sarabjeet Singh
Abstract
Diabetic Foot Ulcer (DFU) is a major global healthcare challenge, entailing substantial costs and increased mortality rates. The aim of this work is to develop a ResNet-50 transfer learning model for DFU detection using thermal images. The ResNet-50 transfer learning model is fine-tuned to adapt its pre-trained features to DFU detection, yielding improved accuracy and dataset generalisation. The developed model achieves 90.52% accuracy, 92.06% precision, 96.97% AUC, and 92.06% recall, showing significant performance gains. Furthermore, a 7% increase in accuracy is observed in comparison to the previous state-of-the-art. This work presents substantial enhancements in DFU detection accuracy, highlighting the potential of deep learning and data augmentation in medical image analysis. The developed model's improved performance underscores its relevance for clinical applications, suggesting its potential to enhance patient care and outcomes.