UKBOB: One Billion MRI Labeled Masks for Generalizable 3D Medical Image Segmentation

2025-04-09

AI Summary

Medical imaging annotation scarcity severely hinders the development of 3D segmentation models. To address this, we introduce UKBOB—the largest open-source MRI organ segmentation dataset to date—comprising 51,000 3D MRI scans, 72 organ classes, and over 1.37 billion 2D instance masks, derived from UK Biobank and curated via automated annotation, organ-specific noise cleaning, and a manually verified subset of 300 cases. We further propose Entropy-Driven Test-Time Adaptation (ETTA) to mitigate label noise during inference. Additionally, we release Swin-BOB, the first foundation model tailored for general-purpose 3D medical image segmentation. Experiments demonstrate that Swin-BOB achieves +0.4% Dice improvement on BraTS and +1.3% on BTCV, while enabling zero-shot cross-modal and cross-domain generalization. All data, code, models, and cleaned labels are publicly released.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Calibration & Uncertainty Quantification
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Large pretrained models with web data
• Security and Privacy: Data transparency and provenance

Abstract

In medical imaging, the primary challenge is collecting large-scale labeled data due to privacy concerns, logistics, and high labeling costs. In this work, we present the UK Biobank Organs and Bones (UKBOB), the largest labeled dataset of body organs, comprising 51,761 MRI 3D samples (equivalent to 17.9 million 2D images) and more than 1.37 billion 2D segmentation masks of 72 organs, all based on the UK Biobank MRI dataset. We utilize automatic labeling, introduce an automated label cleaning pipeline with organ-specific filters, and manually annotate a subset of 300 MRIs with 11 abdominal classes to validate the quality (referred to as UKBOB-manual). This approach allows for scaling up the dataset collection while maintaining confidence in the labels. We further confirm the validity of the labels by demonstrating zero-shot generalization of trained models on the filtered UKBOB to other small labeled datasets from similar domains (e.g., abdominal MRI). To further mitigate the effect of noisy labels, we propose a novel method called Entropy Test-time Adaptation (ETTA) to refine the segmentation output. We use UKBOB to train a foundation model, Swin-BOB, for 3D medical image segmentation based on the Swin-UNetr architecture, achieving state-of-the-art results in several benchmarks in 3D medical imaging, including the BRATS brain MRI tumor challenge (with a 0.4% improvement) and the BTCV abdominal CT scan benchmark (with a 1.3% improvement). The pre-trained models and the code are available at https://emmanuelleb985.github.io/ukbob , and the filtered labels will be made available with the UK Biobank.

Problem

• Addressing limited labeled 3D medical MRI data due to privacy and cost
• Providing large-scale automated organ segmentation with label cleaning
• Enhancing segmentation accuracy via entropy-based test-time adaptation

Innovation

• Automatic labeling with organ-specific filters
• Entropy Test-time Adaptation for noise reduction
• Swin-UNetr based foundation model Swin-BOB