Research Of Cardiac Chamber Segmentation Algorithm In CT Image Based On Stripping

Cardiac segmentation is very important for clinical diagnosis. As it will help doctors forthe diagnosis and treatment of the early of cardiovascular disease by accuratly extracting theregion and the edge of the chamber, then building the3D or4D model. Clinically, doctorsusually have to manually sketch the outline of the heart cavity. Though now somesemi-automatic or fully automatic segmentation method has been used, it still faces manydifficulties in practical application.This paper proposes a segmentation algorithm based onstripping. Firstly, it extracts the heart blood from the complicated background of dual-sourceCT image. Then, each chamber is separated from the heart blood by the algorithm contourstripping and recovery.Finally, it builts a platform for the cardiac segmentation.Based on the mathematical morphology and multi-threshold segmentation method, theheart blood can be extracted. During the segmentation process, because the cardiac image ofdual-source CT, which has been injected contrast agent in imaging procedure, takes thefeature of uneven gray in the same tissue area, so in this paper, the morphologicalreconstruction algorithm is used for filtering and smoothing. Then, used the multi-thresholdsegmentation algorithm to mark the image into different tag according to its regions. Byspecifying the tag which we’d like to extract in the first image, all the same region in differentiamge will be segmented automaticly according to the relationship of same organization in theadjacent images. Consequently, the heart blood in the whole images is segmented bysemi-automaticly.Based the stripping method, it can separates each chamber from the heart blood. Firstly,as the the heart blood is a a shape like "neck", so by stripping the outer contour constantly,the chambers will will be disconnected in the narrow area and remains as cores. Then, bytaking out those cores and recovering their outer contour, the cardiac chambers can berestored. So as to realize the the purpose of segmention of the heart chambers.According to cardiac segmentation algorithm proposed in this paper, a simple platformof cardiac segmentation is set up by using the Qt development toolkit. It mainly includes themodule of image import, isegmentation and display and so on. In the process of platformconstruction, it mainly introduces the method of how to read the infomation of DICOM filesand surface renderingof the three-dimensional data.