Enhancing Diabetic Foot Ulcer Detection and Severity Categorization Using EfficientNet

Diabetic Foot Ulcers (DFUs) are a significant complication of diabetes, necessitating prompt identification and precise evaluation of severity to avert infections, amputations, and other serious consequences. This research utilizes EfficientNet, a cutting-edge convolutional neural network (CNN), to improve diabetic foot ulcer identification and severity classification. The EfficientNet-B0 model was finetuned using a publically accessible dataset of DFU images annotated for ulcer presence and severity, using data augmentation, focus loss, and the Adam optimizer. The model attained an outstanding accuracy of 99.1% in identifying DFUs, with a precision of 98.9%, a recall of 99.2%, and an F1-score of 99.0%. The model achieved a multi-class accuracy of 97.8% for severity categorization, with an AUC of 0.99, indicating exceptional performance across all severity levels. The model's exceptional accuracy and computational economy make it a promising clinical instrument. The findings underscore the capability of EfficientNet in automating DFU identification and severity evaluation, providing a dependable and scalable solution to enhance patient outcomes and alleviate healthcare expenses. Future efforts will focus on augmenting the dataset and improving model interpretability for smooth incorporation into clinical processes.