Detection and Shape Feature Extraction of Breast Tumor in Mammograms

− An accurate and standard techniques for breast tumor segmentation plays a pivotal role in detecting and quantifying breast cancers. Segmentation of breast tumor in mammograms presents many challenges related to selection of optimal threshold in various segmentation techniques. In this paper, a mean based region growing segmentation (MRGS) is presented that automatically finds the seed pixel and optimal threshold value and thus makes the segmentation process very fast and accurate. Furthermore, experimental results are compared with the findings of expert radiologist and marker controlled watershed segmentation approach. A set of 3 mammogram images is used to demonstrate the effectiveness of the segmentation methods. Numerical validation of the results is also provided. Keyword: breast tumor, mean based region growing segmentation, marker controlled watershed segmentation, mammograms, threshold.