Gland segmentation of microscopic image for prostate cancer

Abstract

Computer-aided image processing has attracted much research interest in recent years. In particular, there has been growing interest in processing and analyzing biomedical images by computer. Traditionally, biomedical images used for diagnosis are examined visually by experienced pathologists. However, the subjective nature of visual examination frequently results in variant conclusions by different pathologists. Hence, there's an increasing demand for automated method to help improve the efficiency and accuracy of biomedical images diagnosis by incorporating the power of digital image processing techniques into pathological image analysis. The research work of this thesis focuses on the image segmentation of prostate microscopic images. Prostate cancer is reported as the second leading cause of cancer death in men and the leading cause of cancer death in males over 85. It can only be confirmed by a biopsy, which makes computer-aided image processing an ideal auxiliary tool to help the pathologists make objective diagnosis. This thesis aims to develop a systematic computer-aided method to obtain precise segmentation of abnormal glands from given prostate biopsy images. This research is important and necessary since it constitutes the basis for further computer-aided processing and analyzing of prostate microscopic images. It is easy for human eyes to distinguish the gland from the image due to the color difference between the gland and stroma.