BPD-Neo: An MRI Dataset for Lung-Trachea Segmentation with Clinical Data for Neonatal Bronchopulmonary Dysplasia

2025-06-29

AI Summary

This study addresses the challenge of radiation-free diagnosis of bronchopulmonary dysplasia (BPD) in preterm infants. We constructed the first publicly available, high-resolution 3D MRI dataset of neonatal lungs—comprising 40 free-breathing StarVIBE sequences—accompanied by expert-validated, fine-grained semantic segmentation labels for lungs and trachea, as well as multidimensional clinical metadata. We propose a joint analysis framework integrating 3D stacked star-shaped radial gradient-echo imaging with deep learning, and develop and clinically validate a baseline segmentation model. Key contributions include: (1) the first clinically annotated, open-access 3D MRI dataset for neonatal lung anatomy; (2) radiation-free, high-precision quantification of pulmonary structural morphology; and (3) foundational data and methodology enabling mechanistic investigation of BPD pathophysiology and establishing algorithmic benchmarks for neonatal lung imaging.

Technology Category

• Computer Vision: 3D Computer Vision
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Data Mining & Knowledge Management: Mining of Visual, Multimedia & Multimodal Data

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Bronchopulmonary dysplasia (BPD) is a common complication among preterm neonates, with portable X-ray imaging serving as the standard diagnostic modality in neonatal intensive care units (NICUs). However, lung magnetic resonance imaging (MRI) offers a non-invasive alternative that avoids sedation and radiation while providing detailed insights into the underlying mechanisms of BPD. Leveraging high-resolution 3D MRI data, advanced image processing and semantic segmentation algorithms can be developed to assist clinicians in identifying the etiology of BPD. In this dataset, we present MRI scans paired with corresponding semantic segmentations of the lungs and trachea for 40 neonates, the majority of whom are diagnosed with BPD. The imaging data consist of free-breathing 3D stack-of-stars radial gradient echo acquisitions, known as the StarVIBE series. Additionally, we provide comprehensive clinical data and baseline segmentation models, validated against clinical assessments, to support further research and development in neonatal lung imaging.

Problem

• Develop MRI-based segmentation for neonatal BPD diagnosis
• Provide lung-trachea MRI dataset with clinical annotations
• Enable non-invasive BPD analysis without radiation exposure

Innovation

• High-resolution 3D MRI for neonatal lung imaging
• Semantic segmentation algorithms for lung-trachea analysis
• Free-breathing StarVIBE series for detailed BPD insights