Resumen

LOPEZ-CABRERA, José Daniel; RUIZ-GONZALEZ, Yusely; DIAZ-AMADOR, Roberto  y  TABOADA-CRISPI, Alberto. Fusion Strategies to Automatically Classify Diabetic Foot Ulcer Images using Computer Vision Techniques. Rev cuba cienc informat [online]. 2022, vol.16, n.1, pp.163-179.  Epub 01-Mar-2022. ISSN 2227-1899.

Diabetic foot ulcers are one of the serious complications presented by diabetic patients. The follow-up and identification of the lesions is of vital importance in order to apply a timely treatment because if they are poorly treated, they can lead to the amputation of the limb of the patient and even cause death. This study aims to evaluate different fusion strategies to improve performance rates in the diabetic foot ulcer image classification task. Two fusion approaches are evaluated, at the feature level and at the decision level. Also, two feature selection techniques, ReliefF and MRMR, were used. An SVM classifier with three kernel types was used and combined from five aggregation functions using the best five classifiers. The best results were obtained using the feature-level fusion strategy. These in turn come from the use of classifiers using the feature-level fusion strategy and using feature selection techniques. The results achieved exceed those reported in the literature. Both fusion alternatives coupled with feature selection methods improved automatic classification of diabetic foot ulcer images.

Palabras clave : Computer Vision; Pattern Recognition; Diabetic Foot Ulcers.

        · resumen en Español     · texto en Español     · Español ( pdf )