基于CV模型优化的肝脏MRI图像分割法  

Liver MR Image Segmentation Base on Optimized CV Model

作　　者：李基臣[1] 杨斐[4] 宋殿行 刘坤 Li Jichen;Yang Fei;Song Dianxing;Liu Kun(TCM Hospital of Rizhao,Rizhao,Shandong 276800,China;Department of Image,Rizhao International Heart Hospital,Rizhao,Shandong 276800,China;Department of Equipment,Rizhao Central Hospital,Rizhao,Shandong 276800,China)

机构地区：[1]日照市中医医院设备科,山东日照276800 [2]日照心脏病医院影像科,山东日照276800 [3]日照市中心医院医学装备部,山东日照276800 [4]日照市中医医院,山东日照276800

出　　处：《现代科学仪器》2021年第4期292-294,共3页Modern Scientific Instruments

摘　　要：目的:为解决肝脏病灶边缘变化不明显的磁共振成像(MRI)图像分割问题,提出了一种基于无边缘主动轮廓(CV)模型的优化图像分割方法对病灶区图像进行分割。方法:首先通过传统CV模型与用新边缘函数优化CV模型进行分割,再分别采用Jaccard、Dice系数对图像分割结果定量评价,并对两种模型的分割时间、迭代次数进行分析。结果:传统CV模型的肝脏图像变化不均匀,部分区域存在伪影表现。优化CV模型图像边缘较清晰,无明显伪影;优化CV模型分割时间、迭代次数及Jaccard、Dice系数均低于传统CV模型(P<0.05)。结论:优化后CV模型分割法应用于肝脏MRI图像分割边缘较清晰,分割时间更短,迭代次数减少,分割精确度更高,是理想分割方法,值得临床推广。Objective:To investigate application of optimized Chan-Vese(CV)model in image segmentations of blurry boundary and complex topological structures in liver magnetic resonance imaging(MRI)images.Methods:Firstly,segmentation was performed using the traditional CV model and the optimized CV model with the new edge function,respectively.Then,Jaccard and Dice coefficients were used to quantitatively evaluate the image segmentation results.The segmentation time and iteration times of the two models were also analyzed.Results:The liver image of traditional CV model varied unevenly,and there were artifacts in some areas,while the image of optimized CV model had clear edges and no obvious artifacts;The segmentation time,iteration times and Jaccard and Dice coefficients of optimized CV model were lower than those of traditional CV model(P<0.05).Conclusion:Compared with traditional CV model,application of optimized CV model segmentation method in liver MRI image segmentation has the advantages of clear edge of segmentation,shorter segmentation time,fewer iterations and higher accuracy,which is worthy of clinical promotion.

关 键 词：CV模型优化 肝脏MRI 图像分割 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]