Diabetic foot ulcer detection using interactive image segmentation techniques

Abstract

Diabetic foot ulcer is one of the most dangerous complications of the diabetes and it needs follow-up by the medical doctor. It makes patients life physically more difficult, and insufficient care may lead amputation of the leg. The stage of the ulcer is determined according to Wagner. Segmentation of infection and ulcer through images can help a better understanding of stage of the ulcer. Although there exists automatic segmentation algorithms in the literature,it does not yield good results in complicated cases. For these circumstances medical doctor should want to make segmentation on by placing seeds for corresponding parts, interactive segmentation techniques are proposed. In this study, we used random walker image segmentation algorithm for different number of seeds on 20 diabetic patients. We also have divided the image to superpixels using turbopixels algorithm and applied superpixel based random walker image segmentation algorithm. We have investigated and compared these algorithms.