Self-supervised method improves CT scan analysis with fewer labels
L2GSSL: Label-Efficient Local-to-Global Self-Supervised Learning for Volumetric Healthcare Data
Computer Vision and Pattern RecognitionMachine Learning
Summary
Medical scans contain many images that doctors cannot label fully due to time limits. The authors developed a method called L2GSSL that learns useful features from unlabeled CT scans by looking at parts of each image and the whole volume together. This method helps computers better recognize important patterns, such as lung nodules, even when only a small number of scans are labeled by experts. It performs almost as well as fully supervised methods but requires much less expert annotation.
What this means in practice
- •For hospital data teams: Improve diagnostic models for CT volumes when expert labels are scarce using local-to-global self-supervised pretraining.
- •For medical imaging software developers: Incorporate local and global feature learning from unlabeled 3D scans to boost lesion detection accuracy with fewer annotations.
Authors
Po-Cheng Cheng, Hung-Hsuan Chen
Abstract
Healthcare institutions routinely accumulate high-dimensional computed tomography (CT) volumes, but only a fraction can be annotated by experts. We present L2GSSL, a local-to-global self-supervised learning framework that leverages unlabeled volumes by decomposing pretraining into intra-slice and inter-slice representation spaces. A slice encoder learns view-invariant local features, and a global encoder aggregates the resulting tokens to model volumetric context. The stop-gradient isolates the local space from the global objective, while slice-wise execution and a three-phase schedule support training under constrained GPU memory. On LIDC-IDRI lung nodule malignancy classification, L2GSSL reaches $0.9229$ AUROC with only $10\%$ of the labels and $0.9377$ with $25\%$, exceeding the volume-level VICReg by $0.0229$ and $0.0294$, respectively. With all labels, the strongest supervised baseline exceeds L2GSSL by only $0.0044$ AUROC, consistent with self-supervised pretraining providing its greatest value when annotations are limited. Without adaptation to the external LNDb cohort, a model fine-tuned on $25\%$ of LIDC-IDRI obtains $0.8138$ AUROC and $0.7589$ AUPRC, improving over the strongest evaluated self-supervised baseline by $0.1130$ and $0.0954$. Experiments across data sources, thoracic and abdominal anatomy, downstream tasks, and two backbones examine healthcare-data variety, while lesion-slice masking shows that the predictions depend on lesion-bearing content.