Improved Segmentation of Suspicious Regions of Masses in Mammograms by Watershed Transform

摘要: Abstract Screening mammograms are powerful aids in early detection of breast cancer. This work deals with segmentation suspicious region anomalies known as masses mammogram. The proposed method is based on watershed transform morphologically reconstructed image. mass regions isolated and the results compared two current methods: thresholding graph saliency map preprocessed mammogram morphological extraction from saliency. Quantitative analysis comparison performed using similarity measures classifier performance found that gives better achieving 0.95 measure 83% ROC area.