A comparative study on segmentation methods of micro calcification in mammogram
Bincy Babu, R. J. Hemalatha, A. Josephin Arockia Dhivya, T. R. Thamizhvani, R Chandrasekaran · International Journal of Engineering & Technology · 2018
The primary indication of breast cancer is the presence of calcification clusters. It is challenging and lengthy process for radiologists to identify and classify micro calcifications as non-cancerous or cancerous. In this proposed work, a novel method for the detection of micro calcification clusters in mammograms is explained that consists of two main sections. First, mammogram preprocessing is done. Second, micro calcification are segmented out. In preprocessing noise and label are removed as well as contrast is enhanced. Then various segmentation methods are used for comparison of calcification region. Watershed segmentation, Marker controlled watershed segmentation (MCWS), Texture segmentation and Level set segmentation methods are applied to Digital Database for Screening Mammography (DDSM) database. Results show that the MCWS provides quite acceptable detection performance. The major advantage of this method is its capability to detect micro calcifications perfectly even in case of very dense mammograms. The performance of different methods is evaluated by comparing the obtained segmented image with expert radiologist data. The comparison study aptly shows that the micro calcifications can be exactly segment and can avoid over segmentation problem of existing method.