A Computational Investigation of Breast Tumour on Mammogram Based on Pattern of Grey Scale Distribution

Abstract

Breast cancer is the utmost female tumor and the primary cause of deaths among female. Computer-Aided Detection (CAD) systems are widely used as a tool to detect and classify the abnormalities found in the mammographic images. A detection of breast tumor in a mammogram has been a challenge due to the different intensity distribution which leads to the misdiagnosis of breast cancer. This research proposes a dectection system that is capable to detect the presence of mass tumor from a mammogram image. A total of 160 mammogram images are acquired from Mammographic Image Analysis Society (MIAS) databse, which are 80 normal and 80 abnormal images. The mammogram images are rescaled to 300 x 300 resolution. The noise in the mammogram is suppressed by using a Wiener filter. The images are enhanced by using Power Law (Gamma) Transformation, ɣ = 2 for a better image quality. The greyscale information that contain tumor mass is extracted and used to model the proposed detection system by using 80% or 128 and of the total 160 mammogram images. The rest 20% or 32 mammogram images are used to test the performance of the proposed detection system. The experimental results show that performance of the proposed detection system has 90.93% accuracy.