Department of Radiology, Guangxi Medical University First Affiliated Hospital, No. 6 Shuangyong Rd, Nanning, Guangxi 530021, China.
Huiying Medical Technology Co., Ltd., HaiDian District, Beijing, China.
AJR Am J Roentgenol. 2021 Jun;216(6):1510-1520. doi: 10.2214/AJR.20.23255. Epub 2021 Apr 7.
This study aimed to determine the best model for predicting microvascular invasion (MVI) of hepatocellular carcinoma (HCC) using conventional gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid (gadoxetate disodium)-enhanced MRI features and radiomics signatures with machine learning. This retrospective study included 269 patients with a postoperative pathologic diagnosis of HCC. Gadoxetate disodium-enhanced MRI features were assessed, including T1 relaxation time, tumor margin, tumor size, peritumoral enhancement, peritumoral hypointensity, and ADC. Radiomics models were constructed and validated by machine learning. The least absolute shrinkage and selection operator (LASSO) was used for feature selection, and radiomics-based LASSO models were constructed with six classifiers. Predictive capability was assessed using the ROC AUC. Histologic examination confirmed MVI in 111 (41.3%) of the 269 patients. ADC value, nonsmooth tumor margin, and 20-minute T1 relaxation time showed diagnostic accuracy with AUC values of 0.850, 0.847, and 0.846, respectively ( < .05 for all). A total of 1395 quantitative imaging features were extracted. In the hepatobiliary phase (HBP) model, the support vector machine (SVM), extreme gradient boosting (XGBoost), and logistic regression (LR) classifiers showed greater diagnostic efficiency for predicting MVI, with AUCs of 0.942, 0.938, and 0.936, respectively ( < .05 for all). ADC value, nonsmooth tumor margin, and 20-minute T1 relaxation time show high diagnostic accuracy for predicting MVI. Radiomics signatures with machine learning can further improve the ability to predict MVI and are best modeled during HBP. The SVM, XGBoost, and LR classifiers may serve as potential biomarkers to evaluate MVI.
本研究旨在利用常规钆塞酸二钠（gadoxetate disodium）增强 MRI 特征和放射组学特征与机器学习，确定预测肝细胞癌（HCC）微血管侵犯（MVI）的最佳模型。本回顾性研究纳入了 269 例术后病理诊断为 HCC 的患者。评估了钆塞酸二钠增强 MRI 特征，包括 T1 弛豫时间、肿瘤边缘、肿瘤大小、肿瘤周边强化、肿瘤周边低信号和 ADC。利用机器学习构建并验证放射组学模型。采用最小绝对收缩和选择算子（LASSO）进行特征选择，并利用六种分类器构建基于放射组学的 LASSO 模型。采用 ROC AUC 评估预测能力。组织学检查证实 269 例患者中有 111 例（41.3%）存在 MVI。ADC 值、非光滑肿瘤边缘和 20 分钟 T1 弛豫时间的诊断准确性 AUC 值分别为 0.850、0.847 和 0.846（均<0.05）。共提取了 1395 个定量成像特征。在肝胆期（HBP）模型中，支持向量机（SVM）、极端梯度提升（XGBoost）和逻辑回归（LR）分类器在预测 MVI 方面具有更高的诊断效率，AUC 值分别为 0.942、0.938 和 0.936（均<0.05）。ADC 值、非光滑肿瘤边缘和 20 分钟 T1 弛豫时间对预测 MVI 具有较高的诊断准确性。利用机器学习的放射组学特征可以进一步提高预测 MVI 的能力，在 HBP 中建模效果最佳。SVM、XGBoost 和 LR 分类器可能成为评估 MVI 的潜在生物标志物。
