Related Experiment Video
Updated: Jan 11, 2026
08:41
Novel In Vivo Micro-Computed Tomography Imaging Techniques for Assessing the Progression of Non-Alcoholic Fatty Liver Disease
Published on: March 24, 2023
1.6K
CT-Based Radiomic Models in Biopsy-Proven Liver Fibrosis Staging: Direct Comparison of Segmentation Types and Organ
Andreea Mihaela Morariu-Barb1,2, Tudor Drugan3, Mihai Adrian Socaciu1,2
1Department of Medical Imaging, "Iuliu Hatieganu" University of Medicine and Pharmacy, 400012 Cluj-Napoca, Romania.
Diagnostics (Basel, Switzerland)
|November 13, 2025
Summary
This study shows that 3D CT radiomic models, particularly those combining liver and spleen data, can accurately predict liver fibrosis stages, performing comparably to vibration-controlled transient elastography (VCTE). These models offer a promising non-invasive tool for assessing liver fibrosis.
Area of Science:
- Medical Imaging
- Radiology
- Hepatology
Background:
- Liver fibrosis is a critical prognostic factor in chronic liver diseases (CLD).
- Computed tomography (CT) has limitations in assessing early-stage liver fibrosis.
- Radiomics offers potential for quantitative CT analysis to reveal hidden image patterns.
Purpose of the Study:
- To evaluate automated CT-based radiomic models for predicting biopsy-proven liver fibrosis.
- To compare different segmentation strategies and organ inclusion approaches.
- To assess radiomic model performance against vibration-controlled transient elastography (VCTE) and predict liver steatosis.
Main Methods:
- Retrospective analysis of 58 CLD patients and 9 controls who underwent VCTE and contrast-enhanced CT.
- Extraction of radiomic features using 2D and 3D segmentations of the liver, spleen, and combined liver-spleen.
- Training multilayer perceptron neural networks to predict fibrosis staging (≥F1-F4) and steatosis grading (≥S1-S3).
Main Results:
- 3D radiomic models outperformed 2D models for liver fibrosis prediction.
- The 3D liver-spleen model achieved excellent performance (AUROCs 0.898-0.974), comparable to VCTE (AUROCs 0.909-0.968).
- Radiomic models showed poor predictive ability for steatosis (AUROCs 0.44-0.69) compared to CAP (AUROCs 0.798-0.917).
Conclusions:
- CT-based 3D radiomic models, especially the liver-spleen combination, show significant potential for predicting liver fibrosis stage.
- These models are comparable to VCTE and could be valuable when elastography is unavailable or inconclusive.
- Portal-venous-phase radiomics lack predictive value for steatosis assessment, necessitating further validation in larger studies.
