Image Segmentation Algorithm Based On Cardiovascular And Cerebrovascular Disease Detection

Cardio-cerebrovascular disease(CVD)is the most common disease with the highest mortality and disability rate in clinic.Computed tomography(CT)and digital subtraction angiography(DSA)are commonly used in the detection of CVD.DSA images are often used to display the two-dimensional topological structure of cerebral vessels by angiography.It is an important basis for doctors to diagnose and treat cerebrovascular diseases in hospitals.CT angiograms are often used to display the three-dimensional morphology of cardiovascular diseases,which is an important means of diagnosis and treatment of cardiovascular diseases.Cardiovascular and cerebrovascular segmentation can assist doctors in diagnosis and improve work efficiency and level.Therefore,this paper focuses on the research of segmentation algorithm for the image of cardiovascular and cerebrovascular disease detection.Although image segmentation based on deep learning theory is a hotspot of current research,this method relies on manual sketching to extract blood vessel samples,which is not only time-consuming and laborious,but also difficult to ensure segmentation accuracy.In view of this,this paper focuses on DSA and CT images for vascular segmentation.In blood vessel image,the topological structure and crosssection structure of capillary are similar to the gray-scale pixels of background,and the segmentation error rate is relatively high.In addition,the edge of blood vessel is not smooth in the result of blood vessel segmentation,and the segmentation error is prone to occur at the intersection of blood vessel.Aiming at the above problems,DSA vascular segmentation algorithm based on multi-scale enhancement and CT 3D vascular segmentation algorithm based on graph theory are studied respectively.The main tasks are as follows:(1)In the study of DSA angiography,the commonly used segmentation method is based on multi-scale Hessian matrix,but this method can not solve the problem of vascular blurring and artifact noise introduced in DSA angiography when the soft tissue around the blood vessel contains permeable contrast agent and inadequate illumination conditions.In order to solve this problem,firstly,this paper proposes an algorithm of edge enhancement,which improves the gradient of vessel edge,makes the vessel edge smoother,and enhances the visual effect at the intersection of vessels.Secondly,a morphological noise filtering method is introduced,which can detect and remove linear noise similar to vessels.Finally,for each vessel in DSA sequence.One image can only show part of the blood vessels,which is not conducive to the observation of the overall state of the blood vessels.The fusion of blood vessel sequence images is designed to show the blood vessels in each image on the same image.In this image,the whole state of the blood vessels can be understood.After comparing with the segmentation methods in the relevant references,the improved DSA vascular segmentation method in this paper increases the coincidence rate by 0.0113,and decreases the false segmentation rate by 0.0238.(2)In the three-dimensional vascular segmentation of multi-slice spiral CT images,the method of first segmentation and then three-dimensional reconstruction is adopted.In cross-sectional images,blood vessels are sliced.In order to solve the problem of non-smooth edges and missing micro-vessels in the results of cross-sectional vascular segmentation,a strain-force enhancement filter is improved according to the similar properties of the Hessian matrix parameters of blood vessels and the structural parameters of strain stress in physical mechanics.The edge of blood vessels in crosssectional CT is enhanced to make the edge of blood vessels more smooth and crosssectional.According to the theory of graph cut,the energy value of each edge in the blood vessel graph is calculated by Graph-Cut algorithm,and the blood vessels in the cross-sectional image are segmented at the lowest energy value.The segmented blood vessel sequence is reconstructed by ray projection method to obtain a threedimensional view of the blood vessel that is conducive to observation.Compared with the reference segmentation method,the segmentation accuracy of the blood vessel segmentation method in CT image is improved by 0.238.However,due to the complexity of calculation,the segmentation speed is slightly reduced.