基于MRI直方图模型预测胰腺癌淋巴结转移的初步研究

Histogram Analysis Based on MRI for Predicting Lymph Node Metastasis in Pancreatic Cancer:A Preliminary Study

目的:探讨基于MRI直方图分析术前预测胰腺癌淋巴结转移状态的应用价值。方法:收集经手术病理证实的胰腺癌患者152例,根据淋巴结转移状态将患者分为淋巴结转移组和无淋巴结转移组,并将患者按7∶3的比例分为训练组和验证组。提取MRI图像三维感兴趣区的体积和直方图特征,通过对训练组进行特征选择和建模构建预测模型,并在验证组中进行验证。结果:体积和7个直方图特征被纳入预测模型,模型预测胰腺癌淋巴结转移状态的ROC曲线在训练组中的曲线下面积为0.907,灵敏度为100.0%,特异度为70.9%,在验证组中曲线下面积为0.895,灵敏度为95.5%,特异度为78.3%。结论:基于MRI的直方图模型能有效预测胰腺癌淋巴结转移状态,有助于临床术前评估、术后预后判断。

Purpose:To explore the application value of histogram analysis based on MRI for preoperative predicting lymph node metastasis in pancreatic cancer.Methods:One hundred and fifty-two patients with pancreatic cancer confirmed pathologically were enrolled in this study.The patients were divided into two groups according to the presence or absence of lymph node metastasis,and the patients were divided into training group and validation group according to the ratio of 7∶3.The volume and histogram features were extracted from 3D volume of interest on MRI images,and the prediction model of training group was constructed by feature selection and modeling,the model was further verified in the validation group.Results:The area under the ROC curve of the histogram model was 0.907 in the training group with 100.0%of the sensitivity and 70.9%of the specificity,while the area under the ROC curve for the validation group was 0.895 with 95.5%of the sensitivity and 78.3%of the specificity.Conclusions:MRI based histogram model can effectively predict lymph node metastasis status of pancreatic cancer,which is helpful for preoperative evaluation and postoperative prognosis.