A hybrid approach to diagnosis mammogram breast cancer using an optimally pruned hybrid wavelet kernel-based extreme learning machine with dragonfly optimisation

Abstract

The detection of breast cancer is difficult at the initial stage because the cancerous tumours are rooted in the common breast tissue structures. The objective of this research is to model a breast cancer prediction system with a novel machine learning approach based on wavelets to classify mammogram images as benign, malignant and normal. The prediction of breast cancer for the diagnosis process is made by the proposed algorithm Hybrid Optimally Pruned Wavelet Kernel-based Extreme Learning Machine (HOP-WKELM). Initially, the input is pre-processed for noise reduction using a Kuan filter. Then a quantum evolutionary algorithm (QEA) is applied to segment the cancer part in a mammogram image and extract its features. The extracted features are classified using HWKELM classifier. The HWKELM utilised the dragonfly swarm behaviour-based optimisation (DSBO) approach to optimise the parameters of kernel functions. The proposed strategies achieved a maximum accuracy of 98.8% and maximum precision of 98.1% when compared with AdaBoost systems.