The Study Of Segmentation Algorithm Of Heart CT Image

Heart is an important part of the human body, cardiovascular disease is also increasingly becoming an important threat to human life. Extraction and segmentation of the hearts’ interested portions play a vital role in clinical research. Because it facilitate the doctor’s actions, reduce the errors of human’s operation, improve the efficiency of medical treatment, and save valuable time for doctors and patients.Medical image segmentation is a key steps and important technology of medical image processing, it is also a currently the focus of research .This paper achieves a segmentation for the heart image with the technology of digital image processing.Because Windows cannot directly identify medical images which based on DICOM format. At first, this paper converts the CT image to the image of BMP format which can be identified directly by Windows, this is a basis of the heart image segmentation at following. Then we can study image segmentation after the image has conversed. This paper mainly studies watershed algorithm, single watershed transformation often exists the over-segmentation phenomenon due to the influence of factors such as noise and the inhomogeneous regions. It needs some appropriate improvements. This article preprocess the image of the heart at first, using morphological operations for image filter processing, and then using the improved multi-scale morphological gradient operator to get the image gradient map of the heart.After that acquiring the foreground and background mark by morphological operations and the watershed transformation based on distance, then we modifications the gradient mage to image segmentation with the improved watershed transformation based on tag. Combining segmentation results with the gray characteristics of the target area which is an important prior information, we extract ideal target tissue, then carrying out the appropriate post-treatment based on region merging to obtain the desired separation effect.This topic converts the cardiac CT image which based on DICOM format into the image based on BMP format and display the result of conversion, it is same with the professional software ezDICOM. And this topic can make the batch conversion for the CT images collected from the hospital , greatly improving the efficiency of conversion. Improved watershed algorithm and Combining improved Li level set algorithm with watershed algorithm presented by this paper, can be used to extract the target tissue and remove the adhesion phenomenon. So the result of segmentation is more ideal. Finally the separated target tissues are displayed with pseudo color , it is convenient for observation and analysis.