# Segmentation Pre-training for Label-Efficient Lumbar Spine Degeneration Grading

arXiv: https://arxiv.org/abs/2608.04810 · HTML: https://mlsanity.com/paper/2608.04810/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 7 of its 21 references that are in the map.
- First posted on arXiv 5 Aug 2026.

## Builds on (most cited first, 7 of 7)

- [How Well Do Supervised 3D Models Transfer to Medical Imaging Tasks?](https://mlsanity.com/paper/2501.11253/) (2025)
- [SPINEPS -- Automatic Whole Spine Segmentation of T2-weighted MR images using a Two-Phase Approach to Multi-class Semantic and Instance Segmentation](https://mlsanity.com/paper/2402.16368/) (2024)
- [SpineNetV2: Automated Detection, Labelling and Radiological Grading Of Clinical MR Scans](https://mlsanity.com/paper/2205.01683/) (2022)
- [Relative Attribute Classification with Deep Rank SVM](https://mlsanity.com/paper/2009.07717/) (2020)
- [A Stronger Baseline For Automatic Pfirrmann Grading Of Lumbar Spine MRI Using Deep Learning](https://mlsanity.com/paper/2210.14597/) (2022)
- [End-to-End Diagnosis and Segmentation Learning from Cardiac Magnetic Resonance Imaging](https://mlsanity.com/paper/1810.10117/) (2018)
- [Be Indiscrete: The Benefits of Learning Continuous Spine Degeneration Severity Scores](https://mlsanity.com/paper/2607.05090/) (2026)

## Abstract (arXiv)

Automated assessment of degenerative pathology in the lumbar spine on magnetic resonance imaging (MRI) requires access to large-scale datasets of expert-annotated radiological gradings. In contrast, segmentation pseudo-labels can be generated by automated tools at negligible radiologist cost. We examine whether pre-training on segmentation can effectively replace a fraction of the manual grading annotations required for downstream supervision. We pre-train a 3D ResNet encoder to segment the vertebrae, intervertebral discs (IVDs), and the spinal canal, then fine-tune lightweight task-specific grading heads using different proportions of the available training data, ranging from 10% to 100%. On a multicentre dataset of ~2,000 subjects across 11 pathologies, segmentation pre-training, achieving a Dice score of 0.94 against pseudo-labels, improved the task-averaged (macro) one-vs-rest ROC-AUC at all proportions. With only 20% of grading labels after pre-training, the method achieved near full-supervision performance, with the largest gains observed for either low-prevalence or spatially grounded pathologies.