CT影像中磨玻璃状结节的马尔科夫聚类分割

CT影像中磨玻璃状结节的马尔科夫聚类分割摘要医学图像分割作为对医学图像进行病理分析的基础，它不仅对医学诊断治疗精度有着直接的影响，同时还具有重要的理论意义与实践意义。医学图像分割是通过提取感兴趣的区域，为后续的疾病诊断、治疗计划制定和治疗效果评价提供有价值的参考信息。CT因其分辨力高，能够更加清楚地显示解剖结构及病变组织而被广泛应用于多种系统疾病的诊断。因此，研究其方法在CT图像中的应用具很重要的意义，所以所需要的CT影像处理需求越来越精准，为了减少对病情的误判，对受病者的负责，不得不在CT影像研究更好的方法对其影像对其进行分析：然而，传统的模糊C均值(FCM)聚类分割算法在处理低信噪比图像时，仅考虑了图像的灰度信息，而未充分考虑其空间相关性，从而导致了较大的误差。特别是近年来不断发展的医学检测手段形成大数据，对于图像分割准确性提出更高要求，寻找一种具有更高分割准确性的智能分割算法具有现实意义和很高的应用价值十分必要。针对以上FCM算法的局限性，提出一种基于马尔可夫聚类MarkovClustering,MC)的新分割方法。MC聚类方法相比FCM聚类方法在类内距和类间距的表示上有一定提高，并且适合更加广泛的聚类应用中。从本质上说，MC聚类方法是一种快速且可扩展的无监督图聚类算法，在我们本次的研究中就是将马尔可夫聚类算法运用到图像分割中，利用马尔可聚类的优点将CT图像进行分割，从而达到我们研究的目的。MC聚类方法的验证是通过真实的患者CT影像，通过matlab软件平台实现聚类过程的验证和分析。关键词：CT图像；聚类分割；FCM算法；马尔可夫聚类ABSTRACTAs the basis of pathological analysis of medical images,medical imagesegmentation not only has a direct impact on the accuracy of medical diagnosis andtreatment,but also has important theoretical and practical significance.Medical imagesegmentation provides valuable reference information for subsequent diseasediagnosis,treatment plan formulation and treatment effect evaluation by extractingareas of interest.CT is widely used in the diagnosis of a variety of systemic diseasesbecause of its high resolution,which can show the anatomical structure andpathological tissue more clearly.Therefore,it is of great significance to study theapplication of image segmentation method in CT images,so the required CT imageprocessing needs are more and more accurate.In order to reduce the misjudgment ofthe disease and take responsibility for the patients,better methods have to be used inCT image research to analyze the images.However,the traditional fuzzy C-means(FCM)clustering algorithm only considers the gray information of the image,butdoes not fully consider its spatial correlation when processing low signal-to-noiseratio images,which leads to a large error.Especially in recent years,the continuousdevelopment of medical detection means to form big data,which puts forward higherrequirements for the accuracy of image segmentation,it is necessary to find anintelligent segmentation algorithm with higher accuracy of segmentation has practicalsignificance and high application value.Aiming at the limitations of FCM algorithm,a new segmentation method basedon Markov Clustering (MC)is proposed.Compared with FCM clustering method,MC clustering method has a certain improvement in the representation of intra-classdistance and inter-class distance,and is suitable for more extensive clusteringapplications.In essence,MC clustering method is a fast and scalable unsupervisedgraph clustering algorithm.In this study,Markov clustering algorithm is applied toimage segmentation,and the advantages of Markov clustering are used to segment CTimages,so as to achieve the purpose of our study.The validation of MC clusteringmethod is to verify and analyze the clustering process through real patient CT imagesand matlab software platform.Key words:CT image:Cluster segmentation:FCM algorithm:Markov clustering目录第一章绪论.