摘要
目的探讨基于MRIT2-FLAIR序列的影像组学模型鉴别无水肿型肺腺癌脑转移瘤与腔隙性脑梗死的价值。方法回顾性分析经手术病理或临床与影像随访证实的104例肺腺癌脑转移瘤与165例腔隙性脑梗死患者的治疗前磁共振图像,按DWI序列信号高低分为两组(高信号组105例、等低信号组164例),各组按7∶3随机划分训练集和测试集。从轴位T2-FLAIR序列图像上手动勾画的三维感兴趣区域中提取影像组学特征。采用mRMR和LASSO回归进行降维,筛选出最具诊断价值的影像组学特征,结合4种机器学习分类器分别构建模型,并绘制ROC曲线,采用曲线下面积(AUC)、敏感度和特异度评估各模型的诊断效能。结果等低信号组权重系数最高的特征是一阶特征中的总能量(First Order_Total Energy);高信号组权重系数最高的特征是灰度区域大小矩阵中的小区域低灰度水平强调(GLSZM_Small Area Low Gray Level Emphasis)。4种分类模型中,等低信号组表现最好的是随机森林(RF)模型,其AUC、敏感度和特异度分别为0.887、0.892、0.772(训练集),0.901、0.800、0.939(测试集);高信号组表现最好的是决策树(DT)模型,其AUC、敏感度和特异度分别为0.838、0.892、0.605(训练集),0.816、0.800、0.733(测试集)。结论基于T2-FLAIR序列的影像组学模型对鉴别无水肿型肺腺癌脑转移瘤与腔隙性脑梗死具有一定的价值。
Objective To investigate the value of the image omics model based on MRIT2-FLAIR sequence in differentiating non-edematous lung adenocarcinoma with brain metastases from lacunar infarction.Methods This study retrospectively analyzed the pre-treatment magnetic resonance imaging of 104 cases of lung adenocarcinoma with brain metastases and 165 cases of lacunar infarction confirmed by surgical pathology or clinical observation and imaging follow-up.According to the sequence signal level of DWI,they were divided into two groups(105 cases in the high signal group and 164 cases in the iso-low signal group).Each group was randomly divided into the training set and the test set according to the proportion of 7∶3.Image omics features were extracted from the 3D region of interest that was manually delineated on axial T2-FLAIR sequence images.The mRMR and LASSO regression methods were adopted for dimensionality reduction to screen out the most valuable image omics features.Four machine learning classifiers were combined to construct the models,and the ROC curve was drawn.The area under the curve(AUC),sensitivity and specificity were employed to evaluate the diagnostic effectiveness of each model.Results The feature with the highest weight coefficient in the iso-low signal group was the First Order_Total Energy.The feature with the highest weight coefficient in the high signal group was GLSZM_Small Area Low Gray Level Emphasis.Among the four classification models,the random forest(RF)model had the best performance in the iso-low signal group,and its AUC,sensitivity and specificity were 0.887,0.892,0.772(for the training set),and 0.901,0.800,0.939(for the test set),respectively.The decision tree(DT)model had the best performance in the high signal group,and its AUC,sensitivity and specificity were 0.838,0.892,0.605(for the training set)and 0.816,0.800,0.733(for the test set),respectively.Conclusion The omics model based on T2-FLAIR sequence has certain value in differentiating non-edematous lung adenocarcinoma with brain metastases from lacunar infarction.
作者
汪嫚
俞咏梅
陈鹏飞
Wang Man;Yu Yongmei;Chen Pengfei(Department of Medical Imaging Center,Yijishan Hospital of Wannan Medical College,Wuhu 241001,Anhui,China)
出处
《右江民族医学院学报》
2023年第1期128-133,共6页
Journal of Youjiang Medical University for Nationalities
关键词
磁共振成像
影像组学
脑肿瘤
肿瘤
继发原发性
中风
腔隙性
magnetic resonance imaging
image omics
brain metastases
tumor,secondary primary
apoplexy,lacunar
