Active contour model on digital mammograms using low-cost parallel computing systems

Abstract

Parallel Computing System (PCS) is currently being used widely in many applications of complex problems in image processing. This system will generate the output in shorter periods as compared with sequential computing systems. ARS is a low-cost PCS developed to implement processing of full-field digital mammograms. In this system four processors are used that communicate via Ethernet network using LINUX as the operating system and Matlab Distributed Computing Engine (MDCE) as a platform to process the digital mammograms. The numerical simulation of digital mammograms is based on the Active Contour model in satisfying the finite difference of Euler equation technique. In this paper the performance of the PCS is tested in detecting microcalcifications using a set of mammograms obtained form the Malaysian National Cancer Center (NCC). In this experiment the active contour algorithm is implemented in tracing the icrocalcifications. This paper reports the improvement in time processing, speedup, effectiveness, efficiency and the temporal performance of the system. Results obtained imply that the PCS is prospective for image visualization purposes.