Abstract
Liver fibrosis staging is critical for clinical management, yet automating this process using multi-sequence MRI presents challenges, particularly due to severe class imbalance and limited annotated data. In this paper, we present an end-to-end artificial intelligence framework developed for the MICCAI CARE 2026 challenge (Track 2: CARE-Liver) to simultaneously perform 3D liver segmentation (LiSeg) and fibrosis staging (LiFS). Our dual-task system utilizes four core MRI sequences (GED4, T2, T1, and DWI), effectively managing missing clinical modalities via zero-filling. For LiSeg, we utilize a 3D U-Net trained on 30 ground-truth GED4 masks, serving both as the submission task and the region-of-interest locator for LiFS. To address the significant class imbalance in fibrosis staging, where S4 accounts for approximately 52% of cases, we developed a hybrid ensembled architecture. Our final LiFS model blends a 20-channel 2.5D ResNet34 with a Random Forest classifier trained on 51 handcrafted STAD (Shape, Texture, Appearance, Direction) and vendor features. The framework achieves a validation Dice score of 88.8% for liver segmentation. For fibrosis staging, the blended model yields a validation accuracy of 63.8% and a primary challenge metric score of 0.653. This robust, multi-modality approach demonstrates the efficacy of ensembling deep learning probabilities with radiomic features for challenging, real-world clinical datasets.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/CARE_032.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=lbsAKEdcmz
BibTex
@InProceedings{QiaWei_DualTask_MICCAISAT2026,
author = { Qian, Wei AND Habib, Tushar MD Ahasan AND Fu, Haitao},
title = { { Dual-Task Multi-Sequence MRI Framework: 3D Segmentation and Ensembled Fibrosis Staging of the Liver } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17273},
month = {pending},
page = {pending}
}