An open-access lumbosacral spine MRI dataset with enhanced spinal nerve root structure resolution

Scientific Data, 2024 · DOI: https://doi.org/10.1038/s41597-024-03919-4 · Published: September 23, 2024

Spinal Cord Injury Neurology Medical Imaging

Simple Explanation

This paper introduces an open-access MRI dataset of the lumbosacral spine, focusing on high-resolution imaging of nerve roots to aid in spinal cord neuromodulation research. The dataset includes MRI scans from 14 healthy adults, anatomical annotations of nerve roots, and reconstructed 3D spinal cord models. The dataset aims to address the lack of high-quality spine imaging datasets and promote collaboration in neurorehabilitation engineering.

Study Duration

6 Months

Participants

14 healthy adult volunteers (2 females and 12 males)

Evidence Level

Not specified

Key Findings

1
The dataset provides enhanced resolution of spinal nerve root structures through specific MRI sequences (CISS, DESS, T2-TSE).
2
Image quality was assessed using MRIQC, demonstrating good tissue contrasts and structural details.
3
Annotation differences between two annotators were small enough to not hinder related applications in practice.

Research Summary

The study introduces an open-access lumbosacral spine MRI dataset with enhanced resolution of spinal nerve root structures to facilitate research in spinal cord neuromodulation. The dataset includes MRI scans from 14 healthy adults, anatomical annotations of nerve roots, and reconstructed 3D spinal cord models, assessed for quality using MRIQC. The dataset is expected to promote collaboration among neurorehabilitation engineers and advance research in EES therapy and machine learning algorithms for spine imaging.

Practical Implications

Advancing EES Therapy

The dataset provides a valuable resource for simulating tailored EES therapy due to its detailed insights into human lumbosacral anatomy and spinal cord roots.

Machine Learning Development

The detailed annotations of nerve roots in MRI slices can be used to develop machine learning models for automatic detection and tracing of nerve roots.

3D Spine Model Reconstruction

The reconstructed 3D spine models can be used to develop end-to-end deep learning models to directly reconstruct 3D spine models from MRI data.

Study Limitations

1
The dataset is based on a small sample size of 14 participants.
2
The data were collected from healthy subjects, which may limit the generalizability to patients with spinal cord injuries.
3
Annotation discrepancies highlight the inherent complexity of physiological structures.

Your Feedback

Was this summary helpful?

Back to Spinal Cord Injury