Extreme learning machine for cancer classification in Mammograms based on Fractal and GLCM Features

Abstract

Breast cancer is becoming a leading cause of death among women in the whole world;
meanwhile, it is confirmed that the early detection and accurate diagnosis of this disease can ensure a
long survival of the patients. In this work, classification of Malignant, Benign and normal Images of
Mammogram is performed. Extreme Learning Machine is developed for single hidden layer feedforward
neural network and connection between the input layer and hidden neurons are randomly assigned and
remain unchanged during the learning period. In this work, noises are removed by using Curvelet
transform and Region of Interest (ROI) mammogram images using manual segmentation. By using
Fractal and GLCM features are extracted from the Region of Interest (ROI) and the classification is
performed by using Extreme Learning Machine (ELM) Algorithm. ELM is used to classifies the input
images into normal and abnormal. A set of images (150) from Mammographic Image Analysis Society
(MIAS) database is used for evaluating the system. The performance is analyzed using classification
accuracy rate and reducing the false positive rate. The experimental results are compared with wavelet
transform and Support Vector Machine Method. The result shows that Curvelet Transform with
Extreme Machine Learning (ELM) has high classification accuracy rates of 98.3% than the existing
method for diagnosing the breast cancer. This system will help the radiologist to diagnose the breast
cancer in an efficient way.