Classification of Digital Mammogram Images using Wrapper based Chaotic Crow Search Optimization Algorithm

Abstract

Breast disease is a major cause of death in both women and men around the world. An early diagnosis of breast cancer disease with the assistance of mammogram images is crucial for patients to be treated properly so that they can livea healthy life. While several early-stage strategies for diagnosing breast cancer disease have been developed, most of them are not effective. In this research, an optimized Wrapper-based Chaotic Crow Search Algorithm (WCCSA) is developed to improve the diagnosis of Breast cancer disease. The proposed WCCSA can be used with Probabilistic Neural Network (PNN) to identify the mammogram images asmalignant, benign, and normal, support individuals to receive appropriate care treatment in earlier. The effectiveness of WCCSA with PNNis evaluated using a mini-Mammographic Image Analysis Society (MIAS) dataset of 322 images, and the performance were compared to those of other machine learning algorithms.Anevaluation result shows that the proposed WCCSA with PNN methodseeks an optimal subset of features while maintaining stability, and achieved an accuracy of 97%.