![[2602.19907] Gradient based Severity Labeling for Biomarker Classification in OCT](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.19907] Gradient based Severity Labeling for Biomarker Classification in OCT
Summary
This paper presents a novel strategy for contrastive learning in medical imaging, specifically for classifying biomarkers in OCT scans, improving classification accuracy for Diabetic Retinopathy by 6%.
Why It Matters
The study addresses a critical need in medical imaging by proposing a method that enhances the accuracy of biomarker classification. This is particularly relevant for improving diagnostic tools in ophthalmology, where accurate detection of disease severity can lead to better patient outcomes. By focusing on disease severity characteristics rather than arbitrary augmentations, the approach could set a new standard in medical image analysis.
Key Takeaways
- Introduces a method for generating disease severity labels from OCT scans.
- Improves biomarker classification accuracy by 6% over self-supervised methods.
- Focuses on selecting samples based on disease severity rather than arbitrary augmentations.
- Addresses challenges in medical imaging contrastive learning.
- Potentially enhances diagnostic capabilities for Diabetic Retinopathy.
Computer Science > Computer Vision and Pattern Recognition arXiv:2602.19907 (cs) [Submitted on 23 Feb 2026] Title:Gradient based Severity Labeling for Biomarker Classification in OCT Authors:Kiran Kokilepersaud, Mohit Prabhushankar, Ghassan AlRegib, Stephanie Trejo Corona, Charles Wykoff View a PDF of the paper titled Gradient based Severity Labeling for Biomarker Classification in OCT, by Kiran Kokilepersaud and 4 other authors View PDF HTML (experimental) Abstract:In this paper, we propose a novel selection strategy for contrastive learning for medical images. On natural images, contrastive learning uses augmentations to select positive and negative pairs for the contrastive loss. However, in the medical domain, arbitrary augmentations have the potential to distort small localized regions that contain the biomarkers we are interested in detecting. A more intuitive approach is to select samples with similar disease severity characteristics, since these samples are more likely to have similar structures related to the progression of a disease. To enable this, we introduce a method that generates disease severity labels for unlabeled OCT scans on the basis of gradient responses from an anomaly detection algorithm. These labels are used to train a supervised contrastive learning setup to improve biomarker classification accuracy by as much as 6% above self-supervised baselines for key indicators of Diabetic Retinopathy. Comments: Subjects: Computer Vision and Pattern Recogni...