SpineNet Online Demo
Back pain is the most common cause of long-term disability world-wide, affecting around half of all people over their lifetime. As people live longer, this problem will only get worse.
To combat this, we've developed SpineNet, a computer vision-based system to automatically perform a wide range of radiological gradings in spinal magnetic resonance imaging. It is robust and has been validated across several different datasets, showing performance comparable to clinical radiologists.
You can try a demo version of the software on your own lumbar spine MR scans on this website. We're constantly working on ways to improve it so if you would like to try your own grading system or provide feedback, please contact us.
Try the Software
This is the second version of the spinenet software with better vertebra detection, labelling and additional radiological gradings. To access the original version, click here
The demo version of SpineNet on this website takes lumbar spinal MRIs as input detects and labels vertebral bodies and outputs the following radiological gradings on a per-vertebral disk level:
- Pfirrman Grading (5 Classes)
- Disc Narrowing (4 Classes)
- Central Canal Stenosis (4 Classes)
- Upper & Lower Endplate Defects (Binary)
- Upper & Lower Marrow Changes (Binary)
- Left/Right Foraminal Stenosis (Binary)
- Spondylolisthesis (Binary)
- Disc Herniation (Binary)
We also have versions of the software for detecting and labelling vertebrae in whole spine MRIs and are constantly working on new tasks. Please get in touch if you'd like to try a different task or use the whole spine version.
This software is developed as a collaboration between computer vision researchers and spinal clinicians. We'd particularly like to thank Professor Jeremy Fairbank, Professor Jill Urban, Professor Ian McCall and Dr. Sarim Ather for contributing their expertise in spinal anatomy and pathology to this project.
We are also grateful to Cancer Research UK, EPSRC Program Grant Seebibyte & Genodisc for funding and data acquisition.