文章摘要

文章信息/Info

Title:: Application of image fusion in the classification of liver hydatid disease

作者:: 排孜丽耶?尤山塔依1; 严传波2△; 木拉提?哈米提2; 姚娟3; 阿布都艾尼?库吐鲁克2; 吴淼2; 1.新疆医科大学基础医学院，乌鲁木齐 830011 ；2.新疆医科大学医学工程技术学院，乌鲁木齐 830011；3.新疆医科大学第一附属医院影像中心，乌鲁木齐 830011

Author(s):: PAZILYA?Yusantay1; YAN Chuanbo2; MURAT?Hamit2; YAO Juan3; ABDUGHENI?Kutluk2; WU Miao2; 1.College of Basic Medicine, Xinjiang Medical University , Urumqi 830011, China ; 2.College of Medical Engineering Technology, Xinjiang Medical University , Urumqi 830011;3.Department of Radiology, The First Affiliated Hospital, Xinjiang Medical University,Urumqi 830011

Keywords:: CT image of hepatic hydatid disease; Image fusion; Tamura; Gray-gradient co-occurrence matrix; Image classification

摘要:: 探讨图像融合技术在肝包虫病分型中的应用。对正常肝脏、单囊型肝包虫病、肝囊肿CT图像感兴趣区域分别使用传统的预处理和图像融合方法，对融合后的和预处理后的图像提取Tamura和灰度-梯度共生矩阵特征，通过支持向量机和BP神经网络分类模型进行分类，比较两种方法的分类准确率，并对各分类模型进行参数评估。传统预处理方法对肝囊肿CT图像Tamura和混合特征的分类效果优于图像融合方法，最佳分类准确率为98.333%；图像融合方法对单囊型肝包虫病和正常肝脏CT图像不同特征下的分类准确率均高于传统预处理方法，最佳分类准确率分别为99.167%和100%；图像融合方法不同特征不同分类器下的平均分类准确率高于传统预处理方法。将图像融合方法应用于肝包虫病CT图像的分型中具有一定的分类优势，为肝包虫病影像学诊断提供依据，也为后期研发肝包虫病计算机辅助诊断系统奠定基础。

Abstract:: To discuss application of image fusion technique in the classification of hepatic hydatid disease. Tamura and graygradient cooccurrence matrix textures were extracted from normal liver, singlecystic liver hydatid disease and hepatic cyst CT ROI images used traditional preprocessing and image fusion methods respectively. The classification accuracy of the two methods was compared by using support vector machine and BP neural network classification model, and the parameters of each classification model were evaluated.The traditional preprocessing method was superior to the image fusion method in classifying CT images of hepatic cysts with Tamura and mixed features. The best classification accuracy was 98.333%; the classification accuracy of image fusion for single-cystic liver hydatid disease and normal liver CT images was higher than that of traditional preprocessing method. The best classification accuracy was 99.167% and 100%, respectively; the average classification accuracy of image fusion method with different features and different classifiers was higher than that of traditional preprocessing methods.The application of image fusion method in the classification of hepatic hydatid CT images has certain classification advantages, which provides a basis for the imaging diagnosis of hepatic hydatid and lays a foundation for the later development of the computer-assisted diagnosis system of hepatic hydatid.