An Approach for DFU Classification using VGG-19 and 2D-SSA

Diabetic foot ulcers (DFUs) are a common and severe complication of diabetes, often leading to significant morbidity and potentially limb loss if not properly managed. Early and accurate classification of DFUs is crucial for effective treatment and prevention of further complications. This work focuses on developing an advanced image processing-based system for classifying diabetic foot ulcers to enhance diagnostic accuracy and support clinical decision-making. The system was trained and validated using a dataset of annotated foot ulcer images. The model implemented the VGG-19 model to classify the ulcers based on the extracted features. To further enhance the model efficiency in classification through metrics like accuracy, precision, recall, and F1-score, the input images are reconstructed using 2D singular spectrum analysis (2D-SSA). This 2D-SSA technique uses the eigen vector-based reconstruction retaining the most useful features in the image while discarding the noisy features. Hence, these reconstructed images provided better features to the VGG-19 for a better classification model. The obtained results demonstrated that when VGG-19 was used the reconstructed images gave higher accuracy for classifying DFUs.