The Research And Implementation Of Coronary OCT Images Analysis Algorithm

The cardiovascular and cerebrovascular disease is a very common disease which seriously endangers human health and life.Although the current medical level has been greatly improved.However,nearly 15 million cardiovascular and cerebrovascular disease patients still die each year worldwide.Nearly half of the surviving patients have lost their ability to take care of themselves.Coronary atherosclerotic heart disease(CAD)is a very dangerous and common disease in cardiovascular and cerebrovascular diseases.According to calculations,China has more than 11 million patients with CAD,and the death rate from CAD has surpassed all cancers.Therefore,the prevention and diagnosis of CAD has great significance.Optical coherence tomography(OCT)is a novel intracoronary imaging technique with the advantages of high-resolution,high-resolution,low-impact,real-time imaging.Based on the coronary OCT image,the following research is made for the auxiliary diagnosis of CAD.1.Aiming at segmentation of the plaque area of coronary OCT images,a new algorithm based on K-means clustering and improved random walk was proposed.First,seed points are provided for the random walk algorithm by combining K-means clustering and mathematical morphology,and semi-automated segmentation of different kinds of plaque area is achieved.Secondly,the random walk algorithm is improved by adding the distance between the edge of pixels and the seed point to the definition of the weight function,which effectively reduces the over-segmentation of the weak edge plaque area.2.For the segmentation of the vessel lumen of coronary OCT images,an adaptive level set segmentation model based on shape constraints is proposed.Firstly,the pre-segmentation result of the random walking algorithm is used as the shape constraint of the energy function,The spillover of the evolution curve at the gap of the vessel lumen is prevented.Secondly,taking the growth rate of the mean value of the local entropy outside the evolution curve as an adaptive weight,it effectively combines the CV model and the LBF model,which improves the accuracy of this model for grayscale non-uniform image segmentation and reduces the time taken for segmentation.3.In order to facilitate the use of the hospital and assist clinicians in the diagnosis of coronary heart disease,this thesis designed and implemented the coronary OCT imageanalysis and diagnosis platform.This system is written in Java and Matlab language;MVC framework design;Through the Matlab building for Java interface calls two kinds of algorithms proposed in this article;The entire system is mainly divided into five parts: the user module,the display module,the program interface module,other auxiliary function modules,and the disease auxiliary diagnosis module.Through the debugging of the system,all functions can be realized and run stably.