aiqbench analysis
This research addresses the challenge of multi-organ segmentation in medical imaging by utilizing optimal transport metrics to align feature distributions under partial supervision. The proposed two-stage framework employs learnable organ prototypes and a Sinkhorn-triplet loss to maintain consistency across datasets, effectively mitigating domain shifts without relying solely on pseudo-labels. By explicitly enforcing latent embedding consistency for annotated organs while handling missing annotations, the method offers a computationally efficient alternative that matches state-of-the-art performance on standard benchmarks like BTCV.
Optimal Transport Metric Learning for Feature Alignment in Partially Supervised Segmentation
arXiv:2609.19176v1 Announce Type: cross Abstract: Multi-organ segmentation is often challenged by partially annotated datasets and domain shifts across different imaging sources. To address these limitations, we propose a two-stage learning framework that efficiently leverages partial supervision. In the first stage, the model learns from available annotations to produce accurate segmentations of annotated organs, establishing robust feature representations. In the second stage, we introduce learnable organ prototypes and a Sinkhorn-triplet loss to enforce organ-wise feature consistency across datasets. This encourages latent embeddings of the same organ to remain close, while increasing separation between different organs, even when annotations are missing. Our approach achieves performance comparable to state-of-the-art methods on the BTCV dataset, while remaining computationally efficient. By explicitly aligning feature distributions rather than relying solely on pseudo-labels, the framework effectively mitigates domain shift, making it particularly suitable for medical image segmentation tasks with limited annotation resources.
Abstract:Multi-organ segmentation is often challenged by partially annotated datasets and domain shifts across different imaging sources. To address these limitations, we propose a two-stage learning framework that efficiently leverages partial supervision. In the first stage, the model learns from available annotations to produce accurate segmentations of annotated organs, establishing robust feature representations. In the second stage, we introduce learnable organ prototypes and a Sinkhorn-triplet loss to enforce organ-wise feature consistency across datasets. This encourages latent embeddings of the same organ to remain close, while increasing separation between different organs, even when annotations are missing. Our approach achieves performance comparable to state-of-the-art methods on the BTCV dataset, while remaining computationally efficient. By explicitly aligning feature distributions rather than relying solely on pseudo-labels, the framework effectively mitigates domain shift, making it particularly suitable for medical image segmentation tasks with limited annotation resources.
Image and Video Processing (eess.IV); Artificial Intelligence (cs.AI)
arXiv:2609.19176 [eess.IV]
(or arXiv:2609.19176v1 [eess.IV] for this version)
https://doi.org/10.48550/arXiv.2609.19176 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 237 words
Related coverage
[AINews] Reality Checks on AI News (Yegge shuts down Gas Town, Databricks’ +60% Astra cost)
Latent Space · Sep 17, 2026
Building Blocks for Foundation Model Training and Inference on AWS
Hugging Face · May 11, 2026
Presentation: Engineering AI for Creativity and Curiosity on Mobile
InfoQ · Jul 21, 2026
Fine-tune NVIDIA Nemotron 3 models with Amazon SageMaker AI serverless model customization
AWS · Jul 10, 2026