Large Scale MRI Collection and Segmentation of Cirrhotic Liver

2024-10-06

AI Summary

Automated MRI segmentation and staging of liver cirrhosis are hindered by substantial inter-subject anatomical variability, signal heterogeneity across sequences, and scarcity of high-quality, clinically validated annotations. To address these challenges, we introduce CirrMRI600+, the first large-scale, multi-sequence (T1-weighted/T2-weighted), clinically and histopathologically verified MRI dataset for liver cirrhosis—comprising 628 cases and 39,752 annotated slices, with expert-level pixel-accurate segmentation masks and rigorous quality control. Leveraging this resource, we conduct a systematic benchmark evaluation of 11 state-of-the-art segmentation models (e.g., nnUNet, TransUNet) and propose a novel multi-sequence fusion framework that achieves a mean Dice score of 94.2%. This work establishes the first reproducible, high-fidelity evaluation benchmark for cirrhosis imaging analysis, bridging critical gaps in both standardized data resources and methodological assessment—thereby enabling robust automated visual staging and precision diagnosis.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multi-instance/Multi-view Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Liver cirrhosis represents the end stage of chronic liver disease, characterized by extensive fibrosis and nodular regeneration that significantly increases mortality risk. While magnetic resonance imaging (MRI) offers a non-invasive assessment, accurately segmenting cirrhotic livers presents substantial challenges due to morphological alterations and heterogeneous signal characteristics. Deep learning approaches show promise for automating these tasks, but progress has been limited by the absence of large-scale, annotated datasets. Here, we present CirrMRI600+, the first comprehensive dataset comprising 628 high-resolution abdominal MRI scans (310 T1-weighted and 318 T2-weighted sequences, totaling nearly 40,000 annotated slices) with expert-validated segmentation labels for cirrhotic livers. The dataset includes demographic information, clinical parameters, and histopathological validation where available. Additionally, we provide benchmark results from 11 state-of-the-art deep learning experiments to establish performance standards. CirrMRI600+ enables the development and validation of advanced computational methods for cirrhotic liver analysis, potentially accelerating progress toward automated Cirrhosis visual staging and personalized treatment planning.

Problem

• Accurate segmentation of cirrhotic livers using MRI is challenging
• Lack of large-scale annotated datasets limits deep learning progress
• CirrMRI600+ provides a comprehensive dataset for cirrhotic liver analysis

Innovation

• Large-scale annotated MRI dataset for cirrhotic livers
• Deep learning benchmarks for liver segmentation
• Comprehensive clinical and imaging data integration