Region-growing approach to detect microcalcifications in digital mammograms
Jin‐Wook Shin, Soo‐Ik Chae, Yoon M. Sook, Dong-Sun Park · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2001
Detecting early symptoms of breast cancer is very important to enhance the possibility of cure. There have been active researches to develop computer-aided diagnosis(CAD) systems detecting early symptoms of breast cancer in digital mammograms. An expert or a CAD system can recognize the early symptoms based on microcalcifications appeared in digital mammographic images. Microcalcifications have higher gray value than surrounding regions, so these can be detected by expanding a region from a local maximum. However the resultant image contains unnecessary elements such as noise, holes and valleys. Mathematical morphology is a good solution to delete regions that are affected by the unnecessary elements. In this paper, we present a method that effectively detects microcalcifications in digital mammograms using a combination of local maximum operation and the region-growing operation.