Design And Implementation Of Ventricular Scar Tissue Segmentation System Based On LGE-MRI

At present,the leading cause of death worldwide is ischaemic heart disease,For diagnosis in clinical routine cardiac magnetic resonance imaging is used,as it can provide information on morphology,tissue characterization,blood flow or perfusion.The clinical gold standard for the assessment of myocardial viability is late gadolinium enhanced magnetic resonance imaging(LGE-MRI).The enhancement of the damaged tissue is based on the different contrast agent accumulation within the tissue,which is based on T 1 weighted imaging.Therefore,necrotic tissue has high signal intensity,whereas the boundaries of the myocardium are hardly enhanced.Studies have demonstrated the feasibility of late Gadolinium enhancement(LGE)cardiovascular magnetic,To implement these developed imaging techniques,it is necessary to achieve repetitive and reliable segmentation in the region of myocardial infarction.Thus,this paper is based on the data set with evaluation strategy for the design of ventricular scar tissue segmentation system.Because the LGE-MRI imaging process is disturbed by the potential external noise,the smooth preprocessing step before image segmentation is very important.In this paper,first analyze the gray-scale distribution of the myocardial image file and the myocardial scar image file in the reference image data Feature extraction,and applied to the myocardial image of the Gaussian filter,bilateral filtering and morphological opening and closing based on the noise smoothing technology,obtained by the morphological filtering image is more effective to retain the edge of the image.The data collected in the data center of the myocardial tissue and myocardial scar data,though by many years of clinical experience of the doctors manually segmented,but due to the lack of doctor'pre-knowledge,the orginal image of the intraventricular may wrongly divised.Therefor,this paper realizes the image preprocess based on gray scale correction,and applies the method to the RegionGrow segmentation algorithm.The scores of myocardial scar segmentation before and after pretreatment were quantitatively analyzed.The experimental results show that the gray scale can improve the performance of segmentation algorithm effectively.In this paper,the method of segmentation for system integration mainly includes LevelSet,GraphCuts,WaterShed and regional growth segmentation to realize the segmentation of myocardial scar tissue.The realization of each segmentation algorithm is integrated Different image preprocessing process.In order to evaluate the performance of the segmentation algorithm,this paper uses the method of experiment analysis to analyze the segmentation results of each segmentation algorithm from the Dice index,Jaccard coefficient,Ravd coefficient and the summation rate respectively.The design of the segmentation system mainly includes LGE-MRI image file management,LGE-MRI image noise pretreatment,the segmentation of ventricular scar tissue,three-dimensional imaging based on body rendering method,and quantitative analysis of segmentation algorithm segmentation results And histogram statistical analysis.