MAMMOGRAM IMAGE SEGMENTATION USING WATERSHED

The use of Image Processing techniques in the field of bio-medical imaging is well known. This paper is an attempt to use image processing algorithms for the enhancing and highlighting suspicious areas in mammogram images for breast cancer detection. An image segmentation method based on morphological watershed transform has been developed. It is based on extraction of watershed lines from a topographic representation of the input image. As mammograms are complex images, some kind of pre-processing is required that aids in highlighting suspicious areas. A slight modification to the existing method has been done that help in reducing the number of regions obtained after segmentation. The proposed method has been tested on standard digital mammograms obtained from standard database. The results obtained are quite good and also approved by the radiologists.