New M.E. Thesis Submitted from CSE Student

AUTOMATED DETECTION AND CLASSIFICATION OF MASSES IN MAMMOGRAMS By Gagandeep Singh,cse

Abstract
Breast cancer is the most common cancer in women worldwide. Mammography is a widely used screening tool for detecting subtle sign of malignancy. One of the most common symptom of breast cancer is mass or tumor in breast. A mass can be observed as a bright, hyper-dense object in the breast. The problem lies in segmenting the mammogram to retrieve the mass. Segmentation is choosing the region that highlights the significant properties of the concerned region of Interest (ROI). Despite advances in detection of mass, its segmentation remains to be a hot area of research due to varying shapes and sizes of mass.
Reading masses in mammograms is a very demanding job for radiologists. It is difficult for radiologists to provide accurate result because their judgments depend on training, experience, and subjective criteria.
In the present work, comparison has been performed among the existing segmentation techniques with proposed segmentation technique for detection of masses in mammograms. The proposed techniques works by considering the pixel with the highest intseity in the ROI. The segmentation is performed by drawing certain no of radial lines from the centre towards the boundary of mass with similar angular distance. After that critical points are interpolated to estimate the masses.
Results are presented on the images from the Mammographic image analysis society (mini-MIAS).The proposed algorithm is applied on mammographic images showing the signs of malignancy and benignancy. It clearly classifies the difference between the malignant and benign mass. Results shows that the proposed technique is quite better than state of art techniques used for segmentation of mammographic masses. It acts as a second opinion to assist the radiologist in effective diagnosis of the abnormality due to masses/tumors.



Leisure Readings :