基于表观扩散系数图像的影像组学联合PSA值对前列腺癌侵袭性的诊断价值

The Diagnostic Value of Radiomics Based on Apparent Diffusion Coefficient Image Combined with PSA Value in Prostate Cancer Aggressiveness

目的:研究基于磁共振表观扩散系数(ADC)图像的影像组学联合PSA值对前列腺癌Gleason分级的诊断价值。资料与方法:回顾性分析经手术切除或穿刺活检病理证实的77例前列腺癌患者的临床及影像资料,将患者分为低危组(Gleason评分≤3+4)和高危组(Gleason评分≥4+3)两组。其中高危组57例,低危组20例。基于ADC图像勾画感兴趣区(ROI),测取平均ADC值,另基于ADC图像利用MRIcroGL软件勾画全容积感兴趣区(VOI)提取影像组学特征,经曼惠尼检验及最小绝对值收敛和选择算子(LASSO)降维,结合平均ADC值及临床PSA值,利用Logistic回归建立临床模型,影像组学模型,联合模型(临床-影像组学模型),采用受试者工作曲线(ROC)及曲线下面积(AUC)来评价各种模型的诊断效能。结果:联合模型(临床-影像组学模型)的诊断效能最高(训练组AUC=0.87,测试组AUC=0.91),高于临床模型(训练组AUC=0.76,测试组AUC=0.73)及影像组学模型(训练组AUC=0.85,测试组AUC=0.87)。结论:基于表观扩散系数(ADC)图的影像组学特征对前列腺Gleason分级有诊断价值,且较传统平均ADC值及PSA值的模型表现出更高的诊断效能。

Purpose:To investigate and validate the diagnostic values of the ADC based radiomics combined PSA values in predicting the aggressiveness of prostate cancer.Methods:The ADC maps and clinical factors of 77 patients(20 cases inlow risk group;57 cases in high risk group) with histologically confirmed prostate cancer were analysed retrospectively.The lesions were divided into low risk group(Gleason score≤7) and high risk group(Gleason score>7).Fristly,Region of Interest(ROI) was delineated based on ADC images,and the mean ADC value was measured.In addition,full volume region of Interest(VOI) was delineated based on ADC images using MRIcroGL software,and radiomics features were extracted via Mann-WhitneyU test and least absolute shrinkage and selection operator( LASSO),combined with the mean ADC vlaues and PSA values,then 3 prediction models were established by logistic regression analysis to evaluate the values of these models to predict the aggressiveness of prostate cancer:clinical model,radiomics model and clinical-radiomics combined model.The diagnostic efficiency of each model were compared by receiver operating characteristic(ROC) curve and area under the curve(AUC).Results:The predictive efficacy of the clinical-radiomics combined model in test set were the best(area under curve[AUC]=0.91),The predictive effect of the radiomics model were better(AUC=0.87) than that of clinical model.Conclusion:The radiomics model established by radiomics fetures base on ADC maps could provide important reference in assessment of prostate cancer aggressiveness,and better than the models of mean ADC values combind PSA values.