影像组学对髓母细胞瘤及室管膜瘤诊断价值的研究

Study on the Diagnostic Value of Radiomics in Medulloblastoma and Ependymoma

目的提取髓母细胞瘤(MB)及室管膜瘤(EP)的影像组学特征,探究影像组学对两者的鉴别能力,为早期诊断提供参考。方法回顾性分析诊断为后颅窝内MB及EP患者资料,MB及EP患者各60例,在磁共振T1WI及T2WI序列上绘画感兴趣区,运用3D Slicer共提取特征2072个,运用Mean对数据标准化处理,运用PCC对数据集进行初步降维,运用循环特征消除(RFE)进行特征筛选,运用支持向量机(SVM)进行建模,对训练集进行5次交叉验证,运用准确性、敏感度、特异度、阳性预测值、阴性预测值和95%可信区间对分类器的效能进行评价。结果基于24个影像特征的模型可以获得最高的AUC,其训练集及测试集的AUC值分别达到了0.9983、0.9537,准确性分别为98.81%、88.89%。结论影像组学对鉴别MB及EP具有较高价值,可以作为辅助临床诊断的方法。

Objective To extract the imaging characteristics of medulloblastoma(MB)and ependymoma(EP),and to explore the ability of the radiomics to distinguish the two,so provide reference for early diagnosis.Methods Retrospective analysis was performed on the data of patients diagnosed with MB and EP in the posterior fossa from January 2014 to De⁃cember 2021 in the First Affiliated Hospital of Xinjiang Medical University.There were 60 patients with MB and EP,and the regions of interest were drawn on the T1 WI and T2 WI sequences of the patients.A total of 2072 features were extracted by 3D Slicer.Mean was used for standardized data processing,PCC was used for preliminary dimensionality reduction,Recur⁃sive Feature Elimination(RFE)was used for feature screening,and Support Vector Machine(SVM)was used for model⁃ing.The training set was cross⁃validated for 5 times,and the effectiveness of the classifier was evaluated by using accuracy,sensitivity,specificity,Positive Predictive Value(PPV)and Negative Predictive Value(NPV).Results The model based on 24 image features can obtain the highest AUC,with the AUC values of training set and test set reaching 0.9983 and 0.9537,respectively,and the accuracy is 98.81%and 88.89%,respectively.Conclusion Radiomics has high value in distinguishing medulloblastoma from ependymoma and can be used as an auxiliary method for clinical diagnosis.