基于钆塞酸二钠增强 MRI 的放射组学与临床变量相结合对 HBV 感染患者肝储备功能进行分层。
Gd-EOB-DTPA enhanced MRI based radiomics combined with clinical variables in stratifying hepatic functional reserve in HBV infected patients.
机构信息
Department of Radiology, Northern Jiangsu People's Hospital, Clinical Medical College, Yangzhou University, No. 98 Nantong West Road, Yangzhou, 225001, China.
Department of Radiology, Weill Medical College of Cornell University, 407 E61st Street, New York, NY, 10065, USA.
出版信息
Abdom Radiol (NY). 2024 Apr;49(4):1051-1062. doi: 10.1007/s00261-023-04176-6. Epub 2024 Jan 31.
PURPOSES
To evaluate radiomics from Gd-EOB-DTPA enhanced MR combined with clinical variables for stratifying hepatic functional reserve in hepatitis B virus (HBV) patients.
METHODS
Our study included 279 chronic HBV patients divided 8:2 for training and test cohorts. Radiomics features were extracted from the hepatobiliary phase (HBP) MR images. Radiomics features were selected to construct a Rad-score which was combined with clinical parameters in two models differentiating hepatitis vs. Child-Pugh A and Child-Pugh A vs. B/C. Performances of these stratifying models were compared using area under curve (AUC).
RESULTS
Rad-score alone discriminated hepatitis vs. Child-Pugh A with AUC = 0.890, 0.914 and Child-Pugh A vs. B/C with AUC = 0.862, 0.865 for the training and test cohorts, respectively. Model 1 [Rad-score + clinical parameters for hepatitis vs. Child-Pugh A] showed AUC = 0.978 for the test cohort, which was higher than ALBI [albumin-bilirubin] and MELD [model for end-stage liver disease], with AUCs of 0.716, 0.799, respectively (p < 0.001, < 0.001). Model 2 [Rad-score + clinical parameters for Child-Pugh A vs. B/C] showed AUC of 0.890 in the test cohort, which was similar to ALBI (AUC = 0.908, p = 0.760), and higher than MELD (AUC = 0.709, p = 0.018).
CONCLUSION
Rad-score combined with clinical variables stratifies hepatic functional reserve in HBV patients.
目的
评估钆塞酸二钠增强磁共振成像（MR）的放射组学与临床变量相结合，用于分层乙型肝炎病毒（HBV）患者的肝储备功能。
方法
本研究纳入了 279 例慢性 HBV 患者，将其分为 8:2 的训练组和测试组。从肝胆期（HBP）MR 图像中提取放射组学特征。选择放射组学特征来构建 Rad-score，并将其与两种模型中的临床参数相结合，以区分肝炎与 Child-Pugh A 以及 Child-Pugh A 与 B/C。使用曲线下面积（AUC）比较这些分层模型的性能。
结果
单独的 Rad-score 可区分肝炎与 Child-Pugh A，训练组和测试组的 AUC 分别为 0.890、0.914，Child-Pugh A 与 B/C 的 AUC 分别为 0.862、0.865。模型 1[用于肝炎与 Child-Pugh A 区分的 Rad-score+临床参数]在测试组中的 AUC 为 0.978，高于 ALBI[白蛋白-胆红素]和 MELD[终末期肝病模型]，AUC 分别为 0.716、0.799（p<0.001、<0.001）。模型 2[用于 Child-Pugh A 与 B/C 区分的 Rad-score+临床参数]在测试组中的 AUC 为 0.890，与 ALBI（AUC=0.908，p=0.760）相似，高于 MELD（AUC=0.709，p=0.018）。
结论
Rad-score 结合临床变量可分层 HBV 患者的肝储备功能。
Zotero 插件