A Novel Multi-Modal Approach that Fuses Dermoscopic Images with Thermal Imaging in Pre-Emptive Identification of Diabetic Foot Ulcers (DFUs)

This research presents an innovative multi-modal approach that integrates dermoscopic imaging with thermal imaging data to enhance in pre-emptive identification of DFUs. The proposed methodology leverages the high-resolution visual details of dermoscopic images alongside the temperature distribution data captured by thermal imaging to create a comprehensive diagnostic tool. This fusion of modalities aims to address the limitations of each individual imaging technique—where dermoscopic imaging lacks thermal insights and thermal imaging lacks detailed surface information—thereby improving the early detection and diagnostic accuracy of DFUs.The approach employs an InceptionV3-based CNN combined with a SVM classifier to analyze the fused images. The integration of CNN and SVM enhances the model’s capability to differentiate between healthy skin and early-stage ulcers. Numerous tests using the available datasets showed that the multi-modal fusion greatly improves the classification performance, achieving an accurateness of 97.1% and an overall accuracy of 97.3%. This research contributes to the field by providing a more accurate and early detection method for DFUs, potentially reducing the risk of severe complications such as infection and amputation.