Benchmarking of Deep Learning Methods for Generic MRI Multi-OrganAbdominal Segmentation

2025-07-23

AI Summary

Abdominal multi-organ MRI segmentation suffers from poor model generalizability due to substantial signal intensity variability across scanners and scarcity of high-quality annotated MRI data. Method: We systematically evaluate the zero-shot generalization performance of three state-of-the-art models—MRSegmentator, MRISegmentator-Abdomen, and TotalSegmentator—on cross-vendor, multi-sequence, multi-center MRI data. To alleviate annotation dependency, we propose ABDSynth, a novel synthetic training framework leveraging SynthSeg to generate realistic abdominal MRI-like images and corresponding segmentation labels exclusively from publicly available CT segmentation data—requiring no real MRI annotations. Contribution/Results: MRSegmentator demonstrates superior cross-domain generalizability. ABDSynth achieves practical utility in annotation-scarce settings: while yielding slightly lower accuracy than fully supervised baselines, it significantly reduces reliance on scarce, costly MRI annotations. All code and pretrained models are publicly released.

Technology Category

• Computer Vision: Segmentation
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Recent advances in deep learning have led to robust automated tools for segmentation of abdominal computed tomography (CT). Meanwhile, segmentation of magnetic resonance imaging (MRI) is substantially more challenging due to the inherent signal variability and the increased effort required for annotating training datasets. Hence, existing approaches are trained on limited sets of MRI sequences, which might limit their generalizability. To characterize the landscape of MRI abdominal segmentation tools, we present here a comprehensive benchmarking of the three state-of-the-art and open-source models: MRSegmentator, MRISegmentator-Abdomen, and TotalSegmentator MRI. Since these models are trained using labor-intensive manual annotation cycles, we also introduce and evaluate ABDSynth, a SynthSeg-based model purely trained on widely available CT segmentations (no real images). More generally, we assess accuracy and generalizability by leveraging three public datasets (not seen by any of the evaluated methods during their training), which span all major manufacturers, five MRI sequences, as well as a variety of subject conditions, voxel resolutions, and fields-of-view. Our results reveal that MRSegmentator achieves the best performance and is most generalizable. In contrast, ABDSynth yields slightly less accurate results, but its relaxed requirements in training data make it an alternative when the annotation budget is limited. The evaluation code and datasets are given for future benchmarking at https://github.com/deepakri201/AbdoBench, along with inference code and weights for ABDSynth.

Problem

• Benchmarking deep learning models for MRI abdominal multi-organ segmentation
• Assessing generalizability across diverse MRI sequences and datasets
• Evaluating synthetic training data as an alternative to manual annotations

Innovation

• Benchmarked three MRI segmentation models
• Introduced ABDSynth using CT-based training
• Evaluated generalizability across diverse datasets