Attention Based CNN to Improve Identification of Ischaemia and Infection in DFU

Diabetic foot ulcers (DFUs) frequently occur as complications of diabetes. Identifying infection and ischaemia in DFUs is crucial for wound assessment and treatment planning. Recent advancements in deep learning have shown promise in classifying infection and ischaemia in DFUs using computerized methods. Many modern systems for classifying DFU images rely on deep neural networks, particularly convolutional neural networks. These networks analyze distinctive features and predict class probabilities based on these features. During testing, predictions are made using individual input images and trained parameters. To harness the information in training data more effectively, this study introduces the concept of an attention layer. This layer aids neural networks in memorizing long sequences of data, mitigates issues like exploding gradients, and enhances model accuracy. Experimental results demonstrate that this proposed approach significantly improves the accuracy of DFU infection and ischaemia detection.