双参数磁共振影像组学联合PSAD在前列腺癌Gleason分级分组中的应用价值  

Application value of biparametric magnetic resonance imaging radiomics combined with PSAD in Gleason grade group of prostate carcinoma

作　　者：任大彬 卫雨果 刘丽秋 徐祖良 汪国余 REN Dabin;WEI Yuguo;LIU Liqiu;XU Zuliang;WANG Guoyu(Department of Radiology,Taizhou Central Hospital(Taizhou University Hospital),Taizhou 318000,Zhejiang,China;Advanced Analytics Center,Global Medical Service,GE Healthcare,Hangzhou 310000,Zhejiang,China)

机构地区：[1]台州市中心医院(台州学院附属医院)放射科,浙江台州318000 [2]通用电气医疗集团全球医疗服务部高级分析中心,浙江杭州310000

出　　处：《中国现代医生》2024年第25期30-34,39,共6页China Modern Doctor

基　　金：浙江省台州市第二批社会发展科技计划项目(23ywb29)。

摘　　要：目的探讨双参数磁共振成像(biparametric magnetic resonance imaging,bpMRI)影像组学联合前列腺特异性抗原密度(prostate-specific antigen density,PSAD)在低、高级别前列腺癌(prostate carcinoma,PCa)中的诊断价值。方法回顾性分析2018年6月至2022年10月台州市中心医院经病理证实为PCa患者的临床及影像资料。根据Gleason分级分组(Gleason grade group,GGG),将GGG≤2定义为低级别PCa,GGG>2定义为高级别PCa。按7∶3比例将不同级别的PCa患者随机分为训练组和测试组。基于T2加权成像(T2 weighted imaging,T2WI)、表观扩散系数(apparent diffusion coefficient,ADC)序列提取影像组学特征,采用最大相关最小冗余、最小绝对收缩和选择算子进行特征选择和降维,并进行5倍交叉验证,保留最佳特征组合构建影像组学模型。通过受试者操作特征曲线(receiver operating characteristic curve,ROC曲线)和Delong检验评估各模型的诊断性能。采用决策曲线分析(decision curve analysis,DCA)评价模型的临床效用。结果所有模型中,T2WI-ADC-PSAD联合模型的诊断效能最高,在训练组和测试组中的曲线下面积(area under the curve,AUC)分别为0.882、0.772。Delong检验结果显示,在训练组中,T2WI-ADC-PSAD模型与T2WI模型的AUC比较差异无统计学意义(P>0.05),与其他模型的AUC比较差异均有统计学意义(P<0.05)。在测试组中,T2WI-ADC-PSAD模型与其他模型的AUC比较差异均无统计学意义(P>0.05)。DCA结果显示,当阈值概率低于97%时,T2WI-ADC-PSAD模型可为临床决策提供更高的净效益。结论BpMRI影像组学联合PSAD可提高对低、高级别PCa的诊断效能,并指导患者的治疗决策。Objective To investigate the diagnostic value of biparametric magnetic resonance imaging(bpMRI)radiomics combined with prostate-specific antigen density(PSAD)in predicting low-grade and high-grade prostate carcinoma(PCa).Methods The clinical and imaging data of patients with PCa confirmed by pathology in Taizhou Central Hospital from June 2018 to October 2022 were retrospectively analyzed.According to Gleason grade group(GGG),GGG≤2 was defined as low-grade PCa,and GGG>2 was defined as high-grade PCa.PCa patients with different grades were randomly divided into training group and test group according to a ratio of 7∶3.Radiomics features were extracted based on T2 weighted imaging(T2WI)and apparent diffusion coefficient(ADC)sequences.Feature selection and dimensionality reduction were carried out using maximum relevance minimum redundancy,least absolute shrinkage and selection operator,and 5-fold cross validation was performed to retain the best radiomics features.Receiver operating characteristic(ROC)curve and Delong’s test were used to evaluate the performance of each model.Decision curve analysis(DCA)was used to evaluate the clinical utility of the model.Results Among all the models,T2WI-ADC-PSAD combined model had the best diagnostic efficiency,the area under the curve(AUC)in training group and test group were 0.882,0.772,respectively.Delong’s test showed that in training group,there was no significant difference in AUC between T2WI-ADC-PSAD model and T2WI model(P>0.05),but there were significant differences between T2WI-ADC-PSAD model and other models(P<0.05).In test group there were no significant differences in AUC between T2WI-ADC-PSAD model and other models(P>0.05).The DCA showed that the T2WI-ADC-PSAD model provided a higher net benefit for clinical decision-making when the threshold probability was less than 97%.Conclusion BpMRI radiomics combined with PSAD can improve the diagnostic efficiency of low-grade and high-grade PCa,and guide the treatment decision of patients.

关 键 词：双参数磁共振成像 前列腺特异性抗原密度 影像组学 预测模型 Gleason分级分组 

分 类 号：R737.25[医药卫生—肿瘤]