A Method for Separating Region of InterestIn Digital Mammograms
Sara Dehghani, Amir Masoud Rahmani Mashallah Abbasi Dezfooli · International Journal of Advanced Research in Computer Science · 2011
Tumors in mammography images are appeared as a mass (pile, stack) with high brightness, so darker region and lower gray levels will not need processing. So, first it is better to separate doubtful parts with tumors occurrence probability and limit processing to these parts, but the matter is not in the same easiness since the tumor's background are different. In some cases tumors hide themselves in the denser regions. Sometimes tumors create like a spherical piece in fatty tissues. In some models there is also multiple calcium exist in the tumor, so exact tumor's separation in mammography images is not an easy action. Some people, in normal case, have tumorous breast tissue and these tissues may make mistake with tumors. Since tumorous tissues have high luminosity, when tumors appear in or near these tissues the tumor's margin are less clear and diagnosis is become more difficult. In any case we search for regions with high tumor occurrence and we prefer to separate these parts from mammograms, this work decrease the volume (dimension, size) of mammograms. In this paper we present a method for extract (separate) doubtful region name ROI from mammography images. Our method in this article is in this form that we separate doubtful regions by the usage of image’s energy logarithm, binary imageaccording to the threshold limit's gray level and performing a series of morphology operations. Our method accuracy has been reported to 88% . Key words: Mammography, Cancer Breast Tumor, Region Of Interest, Image Processing, Binary Image, Morphology operation