Automatic Classification of Diabetic Foot Ulcers Using Computer Vision Techniques

Diabetic foot ulcers are one of the common complications that diabetic patients present. Poorly treated lesions can lead to the amputation of the limbs and even cause death. Therefore, the identification and follow-up of the lesions are of vital importance to apply a timely treatment. In this study, we performed the automatic classification of images of diabetic foot ulcers using computer vision techniques. We evaluated different approaches to traditional computer vision techniques and feature extraction from a convolution neural network. An SVM classifier using features extracted by the CNN Densenet201 obtained the best results. The results achieved here outperformed those reported in the literature for similar problems in terms of the F1score measure. That shows that the proposed alternative of combining a pre-trained CNN model as a feature extraction method and then using automatic classifiers is satisfactory in this task.