Department of Diagnostic Radiology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China (Y.X., F.Y., L.L., X.Z.).
Department of Hepatobiliary Surgery, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China (Y.Y., H.Z.); Key Laboratory of Gene Editing Screening and Research and Development (R&D) of Digestive System Tumor Drugs, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China (Y.Y., H.Z.).
Acad Radiol. 2023 Sep;30(9):2010-2020. doi: 10.1016/j.acra.2023.04.030. Epub 2023 Jul 4.
To establish a radiomics nomogram based on multiparameter magnetic resonance (MR) images for preoperatively differentiating intrahepatic mass-forming cholangiocarcinoma (IMCC) from colorectal cancer liver metastasis (CRLM).
A total of 133 patients in training cohort (64 IMCC and 69 CRLM), 57 patients in internal validation cohort (29 IMCC and 28 CRLM), and 51 patients (23 IMCC and 28 CRLM) in external validation cohort were included. Radiomics features were extracted from the multiparameter MR images and selected by the least absolute shrinkage and selection operator algorithm to establish the radiomics model. Clinical variables and magnetic resonance imaging (MRI) findings were selected by univariate and multivariate analyses to construct a clinical model. The radiomics nomogram was combined with radiomics model and clinical model.
Six features were selected to construct the radiomics model. The radiomics signature showed better discrimination than the clinical model in the training cohort (Area Under the Curve (AUC), 0.92; 95% confidence interval (CI), 0.87-0.96 vs. AUC, 0.74; 95% CI, 0.66-0.83) and the external validation cohort (AUC, 0.90; 95% CI, 0.82-0.98 vs. AUC, 0.81; 95% CI, 0.69-0.93). The radiomics nomogram showed the best discrimination performance with favorable calibration in the training cohort (AUC, 0.94; 95% CI, 0.90-0.97) and the external validation cohort (AUC, 0.92; 95% CI, 0.84-1.00).
The radiomics nomogram combining radiomics signatures based on multiparameter MRI with clinical factors (serum carcinoembryonic antigen level and tumor diameter) may provide a reliable and noninvasive tool to discriminate IMCC from CRLM, which could help guide treatment strategies and prognosis preoperatively prediction.
建立基于多参数磁共振成像（MR）的影像组学列线图，用于术前鉴别肝内肿块型胆管细胞癌（IMCC）与结直肠癌肝转移（CRLM）。
纳入训练队列共 133 例患者（64 例 IMCC，69 例 CRLM）、内部验证队列 57 例患者（29 例 IMCC，28 例 CRLM）和外部验证队列 51 例患者（23 例 IMCC，28 例 CRLM）。从多参数 MR 图像中提取影像组学特征，采用最小绝对值收缩和选择算子算法进行选择，建立影像组学模型。采用单因素和多因素分析选择临床变量和磁共振成像（MRI）表现，构建临床模型。将影像组学列线图与影像组学模型和临床模型相结合。
构建影像组学模型时选择了 6 个特征。在训练队列（AUC，0.92；95%置信区间（CI），0.87-0.96 比 AUC，0.74；95%CI，0.66-0.83）和外部验证队列（AUC，0.90；95%CI，0.82-0.98 比 AUC，0.81；95%CI，0.69-0.93）中，影像组学特征的鉴别能力优于临床模型。影像组学列线图在训练队列（AUC，0.94；95%CI，0.90-0.97）和外部验证队列（AUC，0.92；95%CI，0.84-1.00）中表现出最佳的鉴别性能，校准度良好。
基于多参数 MRI 的影像组学特征与临床因素（血清癌胚抗原水平和肿瘤直径）相结合的影像组学列线图可能为鉴别 IMCC 与 CRLM 提供一种可靠、无创的工具，有助于术前指导治疗策略和预测预后。
