Optimum global threshold value detection in mammograms using a phantom

Nowadays, X-ray mammography is the most common method in early detection and diagnosis of breast cancer. In order to shorten the duration of mammogram analysis, and also to contribute radiologists review and analysis, Computer Aided Systems have been developed. One of the first and most important steps of these systems is the segmentation of breast region. In this study, for the breast region segmentation a new derivative-based threshold detection algorithm has been developed. In distinguishing the breast area from the background to determine the most appropriate threshold value, mammogram histograms in MIAS database are examined and classified, phantom mammogram image is prepared and the global threshold detection algorithm is validated.