GUI-CAD Tool to assist the multiclass classification of mammary pathologies by infrared images

Abstract

Infrared thermography is a potential method to improve efficiency for early detection of breast cancer. This technique does not use ionizing radiation and is feasible for screening in men and for detecting changes in young women. In this study, ninety-eight infrared images were used to create a database to develop a computer-aided diagnosis system. Typically, this kind of system is associated with graphical interfaces to facilitate users´ work. In this study, the computer-aided diagnosis was implemented based on statistical classifiers for analysis of four classes: Malignant Tumor, Benign Tumor, Cyst and Healthy. The region of interest was segmented in automatic and semiautomatic ways, which is respectively associated with the Support Vector Machine classifier and Mahalanobis classifier. To evaluate the performance of the proposed classifiers, a confusion matrix was applied to each result obtained. Using the proposed GUI-CAD tool, it was possible to carry out individual and unsupervised classification of patients, with 93% sensitivity.