POPCORN: Progressive Pseudo-labeling with Consistency Regularization and Neighboring

TL;DR

POPCORN is a semi-supervised learning method for medical image segmentation that combines consistency regularization and pseudo-labeling, using a proximity graph to improve label accuracy and reduce bias.

Contribution

It introduces a novel framework that leverages high-level regularization and a proximity graph to enhance pseudo-labeling in medical image segmentation tasks.

Findings

Achieves competitive results on multiple sclerosis lesion segmentation.

Effectively limits confirmation bias through proximity-based data selection.

Outperforms several state-of-the-art SSL strategies.

Abstract

Semi-supervised learning (SSL) uses unlabeled data to compensate for the scarcity of annotated images and the lack of method generalization to unseen domains, two usual problems in medical segmentation tasks. In this work, we propose POPCORN, a novel method combining consistency regularization and pseudo-labeling designed for image segmentation. The proposed framework uses high-level regularization to constrain our segmentation model to use similar latent features for images with similar segmentations. POPCORN estimates a proximity graph to select data from easiest ones to more difficult ones, in order to ensure accurate pseudo-labeling and to limit confirmation bias. Applied to multiple sclerosis lesion segmentation, our method demonstrates competitive results compared to other state-of-the-art SSL strategies.