PLESS: pseudo-label enhancement with spreading scribbles for weakly supervised segmentation

Conference item

Weakly supervised learning with scribble annotations uses sparse user-drawn strokes to indicate segmentation labels on a small subset of pixels. This annotation reduces the cost of dense pixel-wise labeling, but suffers inherently from noisy and incomplete supervision. Recent scribble-based approaches in medical image segmentation address this limitation using pseudo-label-based training; however, the quality of the pseudo-labels remains a key performance limit. We propose PLESS, a generic pseudo-label enhancement strategy which improves reliability and spatial consistency. It builds on a hierarchical partitioning of the image into a hierarchy of spatially coherent regions. PLESS propagates scribble information to refine pseudo-labels within semantically coherent regions. The framework is model-agnostic and easily integrates into existing pseudo-label methods. Experiments on two public cardiac MRI datasets (ACDC and MSCMRseg) across four scribble-supervised algorithms show consistent improvements in segmentation accuracy. Code will be made available on GitHub upon acceptance.

Authors: Gabrielyan, Y; Yeghiazaryan, V; Voiculescu, I

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: IEEE
• Host title: 2026 IEEE International Conference on Image Processing (ICIP)
• Publication date: 2026-08-13
• Acceptance date: 2026-04-30
• Event title: 33rd IEEE International Conference on Image Processing (ICIP 2026)
• Event location: Tampere, Finland
• Event website: https://2026.ieeeicip.org/
• Event start date: 2026-09-13
• Event end date: 2026-09-17
• DOI: 10.1109/ICIP61757.2026.11630101
• EISSN: 2381-8549
• ISSN: 1522-4880
• EISBN: 9798331551513
• ISBN: 9798331551520
• Language: English
• Keywords: image segmentation; weak supervision; scribble supervision; pseudo-labels; scribble spreading; labeling; printing; training; modeling; annotations; dies; biomedical imaging; imaging