Cytological image segmentation based on iterative generalized Hough transform
Zhuofu Liu, Meimei Liu, Lihua Sui, Li Cheng · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2006
Automatic exact segmentation of medical images is very important, since applications need to extract precisely the interesting organic features in the human body. An important example is cell detection in cytological and histological images for the diagnosis of breast cancer. In this paper, we propose a time- and memory-efficient algorithm, called Iterative Generalized Hough Transform (IGHT) for automated cytological image segmentation. In addition to lowering memory requirement, the proposed algorithm reduces the excessive time with image scaling. Instead of being applied to a full-sized image, the IGHT scales down to half-sized and quarter-sized images. The proposed algorithm efficiently exploits both region and edge information. The results show that it is a reliable method for segmenting nuclei in cytological images and for extracting components of interest, which is a key step for diagnosing breast cancer and predicting the course of the disease.