基于钆塞酸二钠增强磁共振成像影像组学定量评估肝硬化患者肝脏储备功能的应用价值

Application Values of Gadolinium-ethoxybenzyl Diethylenetriaminepentaacetic Acid Enhanced Magnetic Resonance Imaging-based Radiomics in the Quantitative Assessment of Liver Reserve Function of Patients with Liver Cirrhosis

目的利用影像组学技术探究基于钆塞酸二钠增强磁共振成像(MRI)肝胆期图像的影像组学特征与肝硬化Child-Pugh分级的相关性,创建列线图预测模型并评价其对肝硬化患者肝脏储备功能的预测价值。方法收集符合纳入标准的100例肝硬化患者,按照Child-Pugh分级标准分成52例Child-Pugh A级患者和48例Child-Pugh B+C级患者,并分别按照7∶3的比例随机分成训练集和测试集。采用AK软件提取训练集患者的钆塞酸二钠增强MRI肝胆期图像的影像组学特征,采用最小绝对收缩与选择算子算法特征选择算法对数据进行降维、特征选择和构建影像组学标签。根据影像组学标签创建预测肝脏储备功能Child-Pugh B+C级的列线图预测模型,将该模型分别应用于训练集和测试集中,并利用受试者工作特征(ROC)曲线定量评估其诊断效能。结果经过对利用AK软件提取的396个纹理特征参数进行降维筛选,最终得到7个影像组学特征参数。根据以上特征建立构建影像组学标签并创建列线图预测模型。通过Wilcoxon秩和检验统计分析训练集和测试集中的Child-Pugh A级和Child-Pugh B+C级的影像组学标签分数值的差异均具有统计学意义(P=0.000,P=0.001)。列线图预测模型预测肝脏储备功能Child-Pugh B+C级的诊断效能在训练集及测试集中的ROC曲线下面积分别为0.88和0.86。结论根据不同肝脏储备功能的肝硬化患者影像组学标签创建的列线图预测模型可以作为肝脏储备功能的较为准确可靠的辅助检测工具,为临床医生更加准确评估肝硬化患者的肝脏储备功能提供一种新手段。

Objective To evaluate the correlation between the radiomics signature of hepatobiliary phase imaging of gadolinium-ethoxybenzyl diethylenetriaminepentaacetic acid(Gd-EOB-DTPA)enhanced magnetic resonance imaging(MRI)and Child-Pugh of liver cirrhosis,establish nomogram prediction model,and assess the predictive value of quantitative assessment of liver reserve function of patients with liver cirrhosis.Methods One hundred patients with liver cirrhosis who met the inclusion criteria were divided into 52 patients with Child-Pugh grade A and 48 patients with Child-Pugh grade B+C according to Child-Pugh classification criteria,and were randomly divided into training set and test set at a proportion of 7∶3.The AK software was used to extract the imaging features of the Gd-EOB-DTPA-enhanced MRI hepatobiliary images of the patients in the training set,and the least absolute shrinkage and selection operator feature selection algorithm was used to reduce the dimension of the data,select the features,and construct the radiomics tags.According to the radiomics label Rad-score,a line chart(nomogram)prediction model was established to predict the Child-Pugh B+C level of liver reserve function.The model was applied to the training set and test set respectively,and the diagnostic efficiency was quantitatively evaluated by receiver operating characteristic(ROC)curve.Results After dimension reduction and screening of 396 texture feature parameters extracted by AK software,7 image feature parameters were obtained.According to the above characteristics,the radiomics tag Rad-score was constructed and the nomogram prediction model was created.The differences of Rad-score scores between Child-Pugh A and Child-Pugh B+C groups in training set and test set were statistically analyzed by Wilcoxon rank sum test(P=0.000,P=0.001).The diagnostic efficacy of nomogram prediction model for predicting Child-Pugh B+C grade of liver reserve function in the ROC curve of training set and test set was 0.88 and 0.86 respectively.Conclusions The nomogram prediction model created according to the radiomics tag Rad-score of patients with liver cirrhosis with different liver reserve functions can be used as a more accurate and reliable auxiliary detection tool for liver reserve function.It provides a new means for clinicians to evaluate liver reserve function more accurately.