The Research On The Key Technology For Stereoscopic Quantitative Assessment On Coronary Artery

Image segmentation technology extract the interest parts of pictures to facilitate the subsequent processing and analysis by collecting image information and using particular methods. Wherein, as the precondition of the computer-aided diagnosis, diseased tissue localization, 3-D reconstruction and treatment program design, the research on medical image segmentation has been highly valued over the years. In the process of stereoscopic quantitative assessment on coronary artery, this paper did the research on coronary extraction and vessel cross-sectional analysis.Firstly, this paper studied the classical image segmentation methods of recent decades, categorizes and summarizes several popular image segmentation methods, briefly described the principles and steps of these methods, analyzed the applicable background and disadvantages places. And described the characteristics of the medical CT images, due to the uneven contrast and organ adhesion, it's a big challenge for CT image segmentation.Secondly, summarized the existing coronary extraction methods, for example, extraction method based on the active contour framework, extraction method based on the vessel tracking and segmentation method based on probabilistic decision. Described the used technical means of these methods, comparing the experimental results, then analyzed the advantages and disadvantages as well as the applicable conditions of these methods. Introduced an coronary analytical model based on hybrid atlas, this model was used to evaluate coronary parameter. The parameter data obtained by the model could be used as a pathological database to provide a reference for future experiments.Thirdly, this paper proposed an extraction method based on three-dimensional morphology of the coronary artery. Study of existing coronary artery extraction methods was tend to focus on the improvement of individual methods, hardly combined with its own characteristics of coronary artery, the researchers always ignored the relationship between the upper and lower pixels during their work on two-dimensional image segmentation. This paper combined histogram threshold, three-dimensional erosion& dilation and other image processing technology, completed automatic identification of the coronary sinus with data analysis,structural features of coronary vascular and relative position properties of the coronary sinus, segmented the pre-split three-dimensional images by the cross section, finally extracted coronary artery and did three-dimensional reconstruction.Finally, this paper proposed a method for vessel profile extraction of coronary artery based on skeleton information. Obtained the coronary artery data by extraction algorithm, selected the appropriate algorithm for skeleton of segmented coronary artery to ensure that skeleton retained the original structural characteristics. Then did smoothing by fitting the skeleton points using Catmull-Rom interpolation algorithm. And calculated tangent vector according to the smoothed skeleton points. Provisioned the mapping relationship between the tangent vector and the three-dimensional coordinate system, determine the profile coordinate system, used the information of tangent vector and the three-dimensional coordinate to accomplish the conversion from profile to three-dimensional coordinates, finally obtained the whole three-dimensional coordinates of profile. Given the size of picture, got the gray value of corresponding points by corresponding relationship, finally finished profile picture after arranged points.This method can achieve a more accurate automatic extraction coronary artery and had a high recognition rate of the coronary sinus. The edges of the automatically segmented coronary artery in all images were distinct and almost matched the original ones. The results also had a good consistency with manual segmentation results. The extraction algorithm proposed by this paper can get accurate cross-sectional images based on the skeleton information. The results can be provided to support the follow-up parameter measurement owing to the rich vessel information and apparent branch condition.