Breast-Cancer Detection using Thermal Images with Marine-Predators-Algorithm Selected Features
V. Rajinikanth, Seifedine Kadry, David Taniar, Robertas Damaševičius, Hafiz Tayyab Rauf
Abstract
Breast-cancer (BC) is one of the major diseases in women group and the early diagnosis and treatment is necessary to cure the disease. Early detection of BC is very essential to implement appropriate treatment and the proposed research aims to develop an automated BC detection system using Breast-Thermal-Images (BTI). The executed approach is as follows; (i) Recording the image for various breast orientation, (ii) Extracting the healthy/DCIS image patches, (iii) Treating the patches with image processing scheme, (iv) Feature extraction, (v) Feature optimization with Marine-Predators-Algorithm (MPA), and (vi) Two-class classification and validation. In this work, the essential image features, such as GLCM and LBP with varied weights are considered to classify the clinically collected BTI into healthy/DCIS class using a chosen two-class classifier. The result of this study confirms that the Decision-Tree (DT) classifier helps to achieve enhanced accuracy (>92%) compared to other methods adopted in this research.