Research On The Segmentation Algorithm Of Coronary Angiography Image

Coronary artery disease is one of the highest incidence rate diseases in cardiovascular disease.It has great threat to human health and life.Therefore,there is an urgent need for prevention and diagnosis of coronary heart disease.Coronary angiography image segmentation is the basis of diagnosis and treatment of coronary heart disease,which can help doctors quickly diagnose the disease.Therefore,the research of coronary angiography image segmentation technology has very important clinical significance.How to improve the segmentation accuracy of coronary angiography image and save doctors’ diagnosis time has always been the research focus of researchers.The gray level of coronary angiography image is uneven,and the shape of blood vessels is changeable,which makes the segmentation of blood vessels more difficult.The existing medical image segmentation technology is difficult to ensure the segmentation accuracy,and is not suitable for all types of vascular images.Therefore,this paper deeply studies the image preprocessing and segmentation technology,adopts the image enhancement method based on bilateral filtering and limited contrast histogram equalization,and combines with different segmentation technologies,designs two kinds of coronary angiography image segmentation methods based on traditional technology and deep learning.The main content are as follows:(1)Coronary angiography image preprocessing.Aiming at the characteristics of background noise and non-uniformity of coronary angiography image,a fusion preprocessing method based on bilateral filtering and limited contrast histogram equalization is proposed.By improving the contrast between vessels and background,the irrelevant noise in the image is suppressed,and the coronary angiography image is enhanced.(2)Coronary angiography image segmentation based on traditional methods.Based on the preprocessing of bilateral filtering and limited contrast histogram equalization,this paper realizes vessel segmentation by multi method fusion.Firstly,according to the different gray values of vessels and background in different regions of the image,the gray image is transformed into a binary image by using adaptive threshold.Secondly,the morphological operation is used to separate the vessels from the noise points and connect the broken points between the vessel segments.Finally,the multi-point region growth is used to extract the vessel structure and realize the image segmentation of coronary angiography.(3)Coronary angiography image segmentation based on deep learning.Based on udensity-net network,a method of coronary artery segmentation is designed,which integrates dense residual blocks and attention mechanism.Firstly,the data set is preprocessed by limited contrast histogram equalization;secondly,the feature expression ability of blood vessel image is improved by cascading dense residual blocks and attention mechanism to the decoder part of u-density-net network,and the local features of blood vessel image are fully extracted to realize the classification of blood vessel and background;finally,morphological operation,threshold segmentation and multi-point region generation are integrated A long post-processing method is used to segment the blood vessel image.Finally,this paper realizes the coronary angiography image segmentation based on traditional methods and deep learning.The algorithm verification and effect evaluation are carried out through the coronary angiography image data set and the results show that the proposed methods in this paper can extract a relatively complete blood vessel structure,and have a better segmentation effect.