Performance Enhancement of Mammogram Preprocessing for Breast Cancer Detection using Improved Successive Mean Quantization Technique

Abstract

In the recent years the rise in cancer cases are at the peak and among different cancer sites the breast cancer is leading the race. It stood second most dangerous cancer among all cancer sites and it is a leading type of cancer among women's and also leading cause of cancer deaths among women's now a days. The thorough studies said that the survival rate of breast cancer is quite good but when it is detected early stage. In this paper we have highlighted the different obstacles that might leads the misprediction by the concern due to many types of artifacts and noise present in the mammogram and it is the main reason of misprediction and accuracy degradation. The paper presents the Improved Successive Mean Quantization Techniques (SMQT) with top hat transform, exponentiation and histogram enhancement to improve the quality of image with accurate region of interest and intensity image enhancement will play to vital role. Proposed model will give better accuracy when tested in MIAS and DDSM dataset it gives approximately 93 to 95 percent accuracy in all mammograms and average overlap ratio is grater that 85% in all tested mammograms.