文章摘要

文章信息/Info

Title:: Morphological quantitative study on liver tumor microvascular network

文章编号:: 1672-6278 (2020)03-0256-10

作者:: 张岁霞1; 2; 王亚勇3; 马燕1; 2; 季学闻2; 4; 邢艳2; 5; 刘慧强1; 2△; 1.新疆医科大学医学工程技术学院，乌鲁木齐 830054;2.省部共建中亚高发病成因与防治国家重点实验室，新疆医科大学第一附属医院，乌鲁木齐 830054;3.新疆医科大学第二附属医院药学部，乌鲁木齐 830054;4.新疆医科大学第一附属医院腹腔镜外科，乌鲁木齐 830054;5.新疆医科大学第一附属医院影像中心，乌鲁木齐 830054

Author(s):: ZHANG Suixia1; 2; WANG Yayong 3; MA Yan1; 2; JI Xuewen2; 4; XING Yan2; 5; LIU Huiqiang1; 2; 1.College of Medical Engineering Technology, Xinjiang Medical University, Urumqi 830054,China;2.State Key Laboratory of Pathogenesis, Prevention Treatment of Central Asian High Incidence Diseases, The First Affiliated Hospital, Xinjiang Medical University, Urumqi 830054;3.Department of Pharmacy, The Second Affiliated Hospital, Xinjiang Medical University, Urumqi 830054;4.Department of Laparoscopic Surgery,The First Affiliated Hospital,Xinjiang Medical University, Urumqi 830054;5.Department of Imaging Center, The First Affiliated Hospital, Xinjiang Medical University,Urumqi 830054

关键词:: 同步辐射相衬显微CT; 肝肿瘤微血管网络; 纹理特征; 特征筛选; 定量分析

Keywords:: Synchrotron radiation phase contrast micro-CT; Liver tumor network structure; Texture features; Feature selection; Quantitative analysis

分类号:: R318

DOI:: 10.19529/j.cnki.1672-6278.2020.03.07

文献标识码:: A

摘要:: 本研究基于同步辐射X射线相衬显微CT的高精度肝肿瘤数据，定量解析肝肿瘤微血管网络结构的灰度和纹理特征变化，探寻其微观病理形态学演化规律。针对肝肿瘤相衬CT三维切片，结合病理分析，提出4种肝肿瘤血管网络演化类型；提取了基于灰度直方图、灰度共生矩阵、灰度-梯度共生矩阵、灰度差分统计及小波能量等49维特征值，构建了肿瘤微血管网络结构特征集。提出曲线下面积（area under curve, AUC）和主成分分析法（principal component analysis, PCA）组合的最优特征筛选法（AUC+PCA），获取能够高效反映图像特征变化的最简特征集；采用Random Forest和决策树C4.5算法挖掘肝肿瘤微血管网络不同演化类型的内涵特征及关联性，并采用参数评估、ROC（receiver operating characteristic，ROC）曲线对分类模型进行定量评价。结果表明，4种肝肿瘤网络结构的内涵特征间均存在组织差异性和分化性，验证了肝肿瘤微损伤分类和演化递进关系。通过肝肿瘤微血管网络形态学特征变化的定量研究具体揭示了肿瘤生长所依赖的炎症微环境与肿瘤浸润发展与转移之间的定量关系，为肿瘤早期检测和治疗策略提供了科学依据。

Abstract:: Based on the high-precision liver tumor data of synchrotron X-ray phase contrast micro-CT, we quantitatively analyzed the changes in the gray and texture features of the microvascular network structure of liver tumors, and explored the development laws of their micropathological morphology. For liver tumor phase contrast CT three-dimensional slices, combined with pathological analysis, we proposed four types of liver tumor vascular network, and extracted 49-dimensional eigenvalues based on the texture features of gray histogram (GH), gray-gradient co-occurrence matrix (GGCM), gray level co-occurrence matrix (GLCM), gray difference statistics (GDS)and wavelet, to construct the feature set of tumor microvascular network structure. The optimal feature selection method of AUC+PCA was proposed. Using Random Forest and C4.5 algorithms to mine the connotative characteristics and correlation of different evolution types of liver tumor microvascular network. We used parameter evaluation and receiver operating characteristic (ROC) curve to evaluate the performance of each classification model quantitatively.The results showed that there were tissue differences and differentiation among the four liver tumor network structures, which verified the classification and evolutionary relationship of liver tumor micro damage. Through the quantitative study of morphological changes of microvascular network in liver tumors, the quantitative relationship between inflammatory microenvironment, tumor invasion, development and metastasis is specifically revealed, which providing scientific basis for early detection and treatment of tumor.

《生物医学工程研究》[ISSN:1006-6977/CN:61-1281/TN]

文章信息/Info

参考文献/References

备注/Memo

常用功能

导航/Navigate

工具/Tools

统计/Statistics