Comparison between mathematical morphology and Canny Edge Detection methods for image post processing techniques in segmenting microcalcification Available to Purchase

Medical image processing is crucial in medical diagnosis, and is mainly used in the early detection of breast cancer. One of the effective imaging techniques in detecting breast abnormalities is screening mammography. However, the size of microcalcification appears too small and very dense, producing very poor contrast anomalies. Besides that, the appearances of inhomogeneous background tissue containing low signal to noise also worsen the detection of microcalcification. Hence, it causes problem for radiologists to extract important information from the image. In order to overcome these problems, an image post processing technique is applied to enhance segmented images for easier interpretation. In this study, Mathematical Morphology and Canny Edge Detection methods are used as post processing techniques. The goal of this study is to choose the best method by comparing the MM and Canny Edge Detection methods in enhancing the boundary extraction of images and to evaluate the accuracy of the boundary extraction of the region in an image. These methods were tested on 20 regions of interest (ROI) images that consist of microcalcification which have been confirmed by radiologists. The relative error between the actual area detected by the radiologists and these two methods were calculated. Experimental result shows that the total percentage error using MM and Canny Edge Detection are 6.02% and 4.77% respectively. This indicates that the MM and Canny Edge Detection methods have successfully segmented the microcalcification of mammogram images with 93.98% and 95.23% accuracy, respectively. Thus, it shows that the Canny Edge Detection method is more accurate for the post processing technique.