Dynamic segmentation of breast tissue in digitized mammograms

Abstract

The percentage of radiodense tissue in a mammogram has been used as a marker for determining breast cancer risk. In this paper, we present an image segmentation technique for identifying tissue and non-tissue regions of a digitized X-ray image. This procedure constitutes a vital step prior to subsequent processing for estimating the amount of radiodense tissue. The process involves the generation of a segmentation mask developed by using discrete wavelet transform techniques. Initial results have been promising, demonstrating the feasibility of the approach.