Breast Cancer Diagnosis from Digital Mammograms Using RF and RF-ELM

Abstract

Artificial neural network is used as emerging diagnostic tool for breast cancer. The objective of this research is to diagnose breast cancer with a machine learning method based on random forest (RF) and RF-ELM classifier. MIAS database is used for digital mammogram images. Preprocessing is generally needed to improve the low quality of image. The region of interest (ROI) is determined according to the size of suspicious area. After the suspicious region is segmented, features are extracted by texture analysis. Gray-level co-occurrence matrix (GLCM) is used as a texture attribute to extract the suspicious area. From all extracted features, best features are selected with the help of correlation-based feature (CBF) selection. Selected features to improve the accuracy of classification are mean, standard deviation, kurtosis, variance, entropy, and correlation coefficient. RF and RF-ELM are used as classifiers. The results of present work show that the CAD system using RF-ELM classifier is very effective and achieves the best result in the diagnosis of breast cancer.