Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Semi-supervised Liver Segmentation and Patch-based Fibrosis Staging with Registration-aided Multi-parametric MRI

Boya Wang, Ruizhe Li, Chao Chen, Xin Chen

TL;DR

This work tackles simultaneous liver segmentation and fibrosis staging from multi-parametric MRI under limited labeled data and cross-modality shifts. It introduces a two-stage framework: BRBS-based semi-supervised segmentation with MRI alignment, and a patch-based fibrosis classifier operating on aligned multi-channel patches, supported by pseudo-labeling and a robust local mutual information loss for registration. The approach yields strong segmentation performance across GED4 and other sequences, and competitive fibrosis staging with good generalization to out-of-distribution data, aided by interpretable patch-level visualizations. Overall, the framework offers a practical, automated pipeline that handles multimodal MRI, reduces reliance on annotated data, and provides clinically meaningful localization of fibrotic changes.

Abstract

Liver fibrosis poses a substantial challenge in clinical practice, emphasizing the necessity for precise liver segmentation and accurate disease staging. Based on the CARE Liver 2025 Track 4 Challenge, this study introduces a multi-task deep learning framework developed for liver segmentation (LiSeg) and liver fibrosis staging (LiFS) using multiparametric MRI. The LiSeg phase addresses the challenge of limited annotated images and the complexities of multi-parametric MRI data by employing a semi-supervised learning model that integrates image segmentation and registration. By leveraging both labeled and unlabeled data, the model overcomes the difficulties introduced by domain shifts and variations across modalities. In the LiFS phase, we employed a patchbased method which allows the visualization of liver fibrosis stages based on the classification outputs. Our approach effectively handles multimodality imaging data, limited labels, and domain shifts. The proposed method has been tested by the challenge organizer on an independent test set that includes in-distribution (ID) and out-of-distribution (OOD) cases using three-channel MRIs (T1, T2, DWI) and seven-channel MRIs (T1, T2, DWI, GED1-GED4). The code is freely available. Github link: https://github.com/mileywang3061/Care-Liver

Semi-supervised Liver Segmentation and Patch-based Fibrosis Staging with Registration-aided Multi-parametric MRI

TL;DR

This work tackles simultaneous liver segmentation and fibrosis staging from multi-parametric MRI under limited labeled data and cross-modality shifts. It introduces a two-stage framework: BRBS-based semi-supervised segmentation with MRI alignment, and a patch-based fibrosis classifier operating on aligned multi-channel patches, supported by pseudo-labeling and a robust local mutual information loss for registration. The approach yields strong segmentation performance across GED4 and other sequences, and competitive fibrosis staging with good generalization to out-of-distribution data, aided by interpretable patch-level visualizations. Overall, the framework offers a practical, automated pipeline that handles multimodal MRI, reduces reliance on annotated data, and provides clinically meaningful localization of fibrotic changes.

Abstract

Liver fibrosis poses a substantial challenge in clinical practice, emphasizing the necessity for precise liver segmentation and accurate disease staging. Based on the CARE Liver 2025 Track 4 Challenge, this study introduces a multi-task deep learning framework developed for liver segmentation (LiSeg) and liver fibrosis staging (LiFS) using multiparametric MRI. The LiSeg phase addresses the challenge of limited annotated images and the complexities of multi-parametric MRI data by employing a semi-supervised learning model that integrates image segmentation and registration. By leveraging both labeled and unlabeled data, the model overcomes the difficulties introduced by domain shifts and variations across modalities. In the LiFS phase, we employed a patchbased method which allows the visualization of liver fibrosis stages based on the classification outputs. Our approach effectively handles multimodality imaging data, limited labels, and domain shifts. The proposed method has been tested by the challenge organizer on an independent test set that includes in-distribution (ID) and out-of-distribution (OOD) cases using three-channel MRIs (T1, T2, DWI) and seven-channel MRIs (T1, T2, DWI, GED1-GED4). The code is freely available. Github link: https://github.com/mileywang3061/Care-Liver
Paper Structure (10 sections, 2 equations, 2 figures, 18 tables)

This paper contains 10 sections, 2 equations, 2 figures, 18 tables.

Table of Contents

  1. Introduction
  2. Methods
  3. Semi-supervised Liver Segmentation and MRI Alignment
  4. Patch-based Liver Fibrosis Staging
  5. Experiments and Results
  6. Dataset
  7. Parameter Setting
  8. Liver Segmentation Results
  9. Liver Fibrosis Classification Results
  10. Conclusions

Figures (2)

  • Figure 1: Overview of the proposed framework. Multi-parametric MRI (DWI, T1, T2, GED1–GED4 with GED4 as reference) are aligned using ANTs tustison2021antsx, and the liver masks of unlabeled GED4 are generated by the BRBS model. Multi-channel patches are extracted by ResNet18 and probability mapping for cirrhosis (S4 vs. S1--3) and substantial (S1 vs. S2--4) fibrosis classification.
  • Figure 2: Visualization of patch-level predictions at different fibrosis stages (Stage 1–4). Blue represents stage 1 patches (pred = 0), while red represents stage 4 patches (pred = 1). The percentage of blue and red patches varies with disease stage, reflecting the progression from early stages to advanced stages.