基于DWI的影像组学在预测弥漫性星形细胞瘤IDH突变中的应用价值  

The Value of DWI Radiomics in Predicting IDH Mutation Status in Diffuse Astrocytoma

作　　者：唐文天 Gobolyos Edit 苏春秋 沈子祯 张久楼 鲁珊珊[1] 洪汛宁[1] TANG Wentian;Gobolyos Edit;SU Chunqiu(Department of Radiology,The First Affiliated Hospital of Nanjing Medical University,Nanjing 210029,P.R.China)

机构地区：[1]南京医科大学第一附属医院放射科,210029

出　　处：《临床放射学杂志》2022年第9期1610-1614,共5页Journal of Clinical Radiology

摘　　要：目的 探讨基于扩散加权成像(DWI)的影像组学在术前无创地预测弥漫性星形细胞瘤异柠檬酸脱氢酶(IDH)突变中的临床应用价值。方法 回顾性分析经手术病理证实的91例弥漫性星形细胞瘤患者(IDH突变型组63例,IDH野生型组28例)的术前常规MRI资料,按8∶2比例随机分为训练组(72例)和验证组(19例)。使用GE A.K软件提取共828个特征,采用主成分分析和最小绝对收缩算子算法进行特征转换和筛选,进而构建影像组学模型,并使用受试者工作曲线下面积(AUC)评估模型的效能。结果 采用逻辑回归、支持向量机和决策树分别构建组学模型,其中逻辑回归模型效能最好,训练组鉴别弥漫性星形细胞瘤IDH突变型与野生型的AUC为0.865,准确率为0.833,敏感度和特异度分别为88.0%、72.7%;验证组AUC为0.833,准确率为0.737,敏感度和特异度分别为84.6%、50.0%。结论 基于DWI的影像组学模型有助于在术前无创地预测弥漫性星形细胞瘤IDH突变。Objective To explore the value of using radiomics based on diffusion weighted imaging(DWI) in predicting isocitrate dehydrogenase(IDH) gene mutation in diffuse astrocytoma. Methods 91 patients pathologically confirmed gradeⅡ astrocytoma(IDH mutant 63 cases;IDH wild type 28 cases) who underwent preoperative conventional MRI examination were enrolled in this retrospective study.The cases were divided randomly into a training cohort(n=72) and a validation cohort(n=19) at a ratio of 8∶2. 828 radiomic features were extracted by A.K(GE Healthcare) software.Principal component analysis(PCA) and least absolute shrinkage and selection operator(LASSO) were used to transform and select features.Radiomics models were finally conducted to obtain the optimized subset of the features, and area under the curve(AUC) was performed to evaluate the performance of the radiomics model. Results Logistic regression(LR),support vector machine(SVM),and decision(DT) were used to conduct the radiomics models, respectively.The LR model showed the best performances in both the training dataset(AUC value was 0.865,accuracy rate was 0.833,sensitivity and specificity were 88.0% and 72.7%) and the validation dataset(AUC value was 0.833,accuracy rate was 0.737,sensitivity and specificity were 84.6% and 50.0%). Conclusion Radiomics models based on DWI are helpful to non-invasively predict IDH gene mutation in diffuse astrocytoma before surgery.

关 键 词：弥漫性星形细胞瘤 异柠檬酸脱氢酶 影像组学 扩散加权成像 磁共振成像 

分 类 号：R445.2[医药卫生—影像医学与核医学] R739.4[医药卫生—诊断学]