结合MRI纹理与SVM的子宫内膜癌肌层浸润程度预测  

MRI Texture Features Combined with SVM to Predict the Depth of Myometrial Invasion in Endometrial Cancer

作　　者：朱雪亮 应捷[1] 杨海马[1] 李薄羏 ZHU Xue-liang;YING Jie;YANG Hai-ma;LI Bo-yang(School of Optical-Electrical and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China)

机构地区：[1]上海理工大学光电信息与计算机工程学院,上海200093

出　　处：《软件导刊》2022年第1期108-113,共6页Software Guide

基　　金：国家自然科学基金项目(U1831133);上海理工大学医工项目(2020)。

摘　　要：子宫肌层浸润深度直接关系到子宫内膜癌患者的治疗和预后,通常使用磁共振成像进行人工评估,受主观影响较大。基于磁共振图像,提出一种计算机辅助判断子宫肌层浸润深度的方法。该方法仅需要计算机或人工提供较容易辨识的子宫体区域就能自动估计浸润程度。首先基于Otsu和形态学处理分割病灶区域,然后提取并融合病灶区域的一阶纹理特征和灰度共生矩阵特征,最后训练支持向量机进行浸润程度分类。实验结果表明,该方法的accuracy达到86.1%、sensitivity达到68.4%、specificity达到91.7%,优于常见分类器,对于辅助判断肌层浸润程度具可行性,且有助于未来从病灶和子宫肌层提取并融合更多种类的特征以提高分类性能。The depth of myometrial invasion(MI)affects the treatment and prognosis of patients with endometrial cancer,commonly evaluated using MRI,which is greatly influenced by subjective factors.Based on MRI,propose a computer-aided diagnosis method for the depth of MI.This method only requires the corpus uteri region provided by humans or computers as input,which is easier to identi⁃fy,and then it estimates the depth of MI automatically.First,the tumor region is segmented based on Otsu and morphological process⁃ing.Then the first order texture features and GLCM features are extracted.Finally,SVM is trained for the depth of MI classification.This method achieved an accuracy of 86.1%,sensitivity of 68.4%and specificity of 91.7%,which outperformed the commonly used classifiers.The results show that the proposed method is feasible for the auxiliary determination of the depth of MI and helpful to extract and fuse more kinds of features from tumor and myometrium to improve the classification performance in future work.

关 键 词：SVM 特征提取 肌层浸润程度 MR图像 计算机辅助诊断 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]