We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
LI-RADS category 3, 4, and M observations: a multiple parameters diagnostic model for hepatocellular carcinoma.
- Authors
Chen, Jianwei; Chen, Huizhen; Zheng, Dechun; Yan, Chuan; Ye, Rongping; Wen, Liting; Li, Yueming
- Abstract
Background: Hepatic lesions categorized as LR-3, LR-4, and LR-M are challenging to accurately assess and diagnose. Purpose: To combine potential clinical and/or magnetic resonance imaging (MRI) features for a more comprehensive hepatocellular carcinoma (HCC) versus non-HCC diagnosis for patients with LR-3, LR-4, and LR-M graded lesions. Methods: Data were consecutively retrieved from 82 at-risk patients with LR-3 (n = 43), LR-4 (n = 20), and LR-M (n = 23) lesions. Significant findings for the differentiation of HCC and non-HCC, including MRI features and clinical factors, were identified with univariable and multivariable analyses. The variables for a prediction model were selected through stepwise use of Akaike's Information Criterion (AIC) to build multivariable logistic regression model. Results: Serum alpha-fetoprotein (AFP) >16.2 ng/mL (odds ratio [OR] = 22.4; P = 0.006), septum (OR = 52.1; P = 0.011), and hepatobiliary phase (HBP) hypointensity (OR = 40.2; P = 0.001) were confirmed as independent predictors of HCC. When combining the three predictors and mild-moderate T2 hyperintensity, the model (AIC = 50.91) showed good accuracy with a C-index of 0.948. Conclusion: In at-risk patients with LR-3, LR-4, or LR-M lesions, integrating AFP, septum, HBP hypointensity, and mild-moderate T2 hyperintensity achieved high diagnostic performance for the diagnosis of HCC.
- Subjects
MAGNETIC resonance imaging; AKAIKE information criterion; LOGISTIC regression analysis; ODDS ratio; REGRESSION analysis
- Publication
Acta Radiologica, 2023, Vol 64, Issue 12, p2977
- ISSN
0284-1851
- Publication type
Article
- DOI
10.1177/02841851231203830