Diabetic Foot Ulcer Diagnosis using Machine Learning

Diabetic Foot Ulcers (DFUs) are a serious diabetes complication, which mostly results in infection, amputation, and increased mortality if not addressed timely. Proper classification of DFUs is critical for early treatment and improved clinical outcomes. In the context of DFU image classification, this study compares the effectiveness of a Convolutional Neural Network (CNN) with traditional Machine Learning (ML) models Random Forest (RF), Support Vector Machine (SVM), and K-Nearest Neighbours (KNN). The CNN model is shown to outperform other models with much higher accuracy since it can extract pertinent features automatically from the images, while ML models depend on manually designed features and have comparatively poorer performance. The research also points to some major gaps in DFU classification, such as excessive intra-class variability, small dataset size, and class imbalance. The results have real-world applicability in the healthcare field by yielding AI-based DFU classification for timely diagnosis and intervention. Future work can build on improving model performance by increasing the dataset size, using transfer learning, and real-time deployment in clinics. The proposed model achieved 90.5% accuracy, 87.2% precision, 85.8% recall, and an F1-score of 86.78%.