Research On Automatic Recognition Algorithm Of Coronary Vulnerable Plaque From OCT Image

Coronary heart disease is one of the world's highest mortality rates,with nearly 70% of acute coronary disease caused by rupture of coronary arterioles.The main reason for medically recognizing vulnerable plaques is macrophage infiltration,so identification of macrophages is an important means of early diagnosis of vulnerable plaques.Clinically,only through the optical coherence tomography observed in patients with coronary macrophages,the doctor by virtue of experience to determine the location of macrophages and the amount of the number of low efficiency and reproducibility is poor.In this paper,the recognition of macrophages in OCT images of coronary arteries,focusing on automatic identification algorithm,and strive to improve the image recognition speed and accuracy,to meet the clinical needs.Firstly,aiming at the study of macrophage recognition algorithm and the characteristics of clinical OCT images,this paper mainly uses the threshold segmentation method and mathematical morphology method to preprocess the original image,and retains the useful information to remove the noise.Secondly,according to the macrophage recognition method needed by clinical diagnosis,three kinds of algorithms are used to identify the macrophages in the image.The first method is based on the gray value of the recognition algorithm,the use of macrophage gray value was significantly lower than the characteristics of normal tissue,using k-means clustering algorithm in the image enhanced to show the edge of macrophage distribution.The second method is based on the Gaussian function fitting recognition algorithm,the use of Gaussian function on the OCT A-scan to fit,combined with the characteristics of the parameters and image classification,and in the image shows the location of macrophages.The third method kind is the recognition algorithm based on texture analysis.The traditional texture analysis method is applied to macrophage recognition,and the image is classified and identified with support vector machine.An improved recognition algorithm is proposed for the deficiency of traditional texture analysis method.It can get better test results.Finally,this paper makes a comparative analysis of the three macrophage recognition algorithms used in this paper.In order to ensure the use of the algorithm,the display interface is prepared,and the images can be processed by interactive operation to facilitate the clinical application.