Breast Cancer Detection in Thermal Images Using GLRLM Algorithm

Abstract

In recent years it has been noticed that early breast cancer detection can decrease death rates considerably and to pursue early detection, there is a need for advanced screening tool along with experts, among screening tools infrared camera in thermography is low cost, contactless and does not include vulnerable rays, so it can be a good alternative to the most common screening tool techniques like mammography which entails all of the mentioned limitations. This paper aims to introduce an architecture by which the computer automatically classifies the cases into the malignant, benign and normal using labeled Thermal breast images. To obtain our goal, Gray Level Run Length Matrix (GLRLM) algorithm for feature selection and Long Short-Term Memory (LSTM) as a classifier are utilized. We achieved near 100% accuracy result for the training process, and for testing, we are selecting eight trained images of a single patient and we get quite accurate outcome. This proposed method using thermal images is a completely non-invasive method for cancerous patients in comparison to other methods.