Abstract
Objective This study aims to evaluate the accuracy of a three-phase T1-derived model in staging liver fibrosis in patients with chronic hepatitis, in comparison with magnetic resonance elastography (MRE). Methods A total of 174 patients with chronic hepatitis were enrolled. The acquisition of three-phase longitudinal relaxation times (T1-Unenh = A, T1-DLP = B, T1HBP = C) from magnetic resonance enhancement examinations, as well as magnetic resonance elastography (MRE) and liver biopsy information for patients.Binary logistic regression was employed to construct models ABC. The diagnostic performance of these models was evaluated using the receiver operating characteristic (ROC) curve and DeLong’s test. Furthermore, the MRE and ABC models were subjected to a comprehensive analysis through 10-fold cross-validation. Results In all liver fibrosis stages (≥ F1-≥F4), both MRE and the ABC model showed moderate correlation with METAVIR fibrosis staging, with significant differences between groups (all P < 0.05). The area under the curve (AUC) for each group of the ABC model was above 0.84. DeLong’s test indicated that in the ≥ F2, ≥F3, and ≥ F4 groups, the ABC model was comparable to MRE (all P > 0.05). Ten-fold cross-validation further confirmed that only in the ≥ F3 group did the ABC model outperform MRE in terms of comprehensive performance. Conclusion This study successfully validated the effectiveness of the three-phase T1-derived model in staging liver fibrosis in patients with chronic hepatitis. In stages ≥ F2, ≥F3, and ≥ F4, it is comparable to MRE, especially showing superior application value in the ≥ F3 stage.
Funding Information
  • Hospital National Natural Science Foundation Cultivation Project (2021GZRPYMS06)
  • Five-Five project of the Third Affiliated Hospital of Sun Yat-sen University (2023WW605)

This publication has 34 references indexed in Scilit: