基于磁共振灌注加权成像和支持向量机鉴别脑泡型包虫病与脑转移瘤

Differential diagnosis of cerebral alveolar echinococcosis and brain metastases based on MRI perfusion weighted imaging and support vector machine

目的探讨磁共振灌注加权成像(perfusion weighted imaging,PWI)及支持向量机(support vector machine,SVM)在脑泡型包虫病(cerebral alveolar echinococcosis,CAE)和脑转移瘤(brain metastase,BMT)鉴别诊断的效能。材料与方法回顾性收集经病理或临床诊断为CAE与BMT的病例各15例,测量两组病灶实性区、灶周水肿和对侧相对正常实质灌注参数,包括相对脑血容量(relative cerebral blood volume,rCBV)、相对脑血流量(relative cerebral blood flow,rCBF)、平均通过时间(mean transit time,MTT)和达峰时间(time to peak,TTP),评估各参数诊断CAE与BMT的效能,并基于灌注参数,运用SVM等机器学习方法鉴别两种疾病。结果CAE病灶实性区rCBF,rCBV诊断曲线下面积(area under curve,AUC)为0.739,0.710,BMT病灶实性区rCBF,rCBV诊断AUC为0.960,0.913,CAE与BMT病灶实性区rCBF和rCBV的诊断效率高于MTT和TTP;CAE与BMT病灶实性区rCBF、rCBV、TTP值差异有统计学意义(P<0.01)。CAE水肿区与BMT水肿区rCBF、rCBV值差异有统计学意义(P<0.01);基于病灶实性区灌注参数,运用SVM分类器可提高鉴别准确率。结论PWI可为CAE与BMT的鉴别诊断提供客观依据。SVM分类方法可提高PWI鉴别两种病灶的准确率。

Objective:To investigate the efficacy of MRI perfusion weighted imaging(PWI)and support vector machine(SVM)in the differential diagnosis of cerebral alveolar echinococcosis(CAE)and brain metastases(BMT).Materials and Methods:The records of patients who pathologically or clinically diagnosed with CAE(15 patients)and BMT(15 patients)were reviewed retrospectively.The perfusion parameters(cerebral blood flow,cerebral blood volume,mean transit time and time to peak)of solid area,perilesion edema and contralateral relative normal area in both lesions were measured and used for evaluation of CAE and BMT.Based on perfusion parameters,SVM and other classifiers were used to identify two diseases.Results:In the diagnosis of CAE/BMT,r CBF and r CBV of solid area achieved area under curve(AUC)of 0.739/0.960 and 0.710/0.913.The diagnostic efficiency of r CBF and r CBV in solid area was higher than that of MTT and TTP;The r CBF,r CBV value of CAE solid area were significantly lower than that of BMT and TTP value was significantly higher(P<0.01).The r CBF,r CBV value of CAE prelesion edema were significantly lower than that of BMT(P<0.01);Based on the perfusion parameters of lesion solid area,the use of SVM classifier can improve the discrimination accuracy.Conclusions:PWI can provide an objective basis for the differential diagnosis of CAE and BMT.SVM classifier can enhance the accuracy of PWI in differentiation of these two diseases.