SCKAN: Structural Consensus-based KAN Prototype Learning for Semi-Supervised Pancreas Segmentation

Researchers introduced SCKAN, a semi-supervised pancreas segmentation method for medical image analysis.

The model combines structure-constrained prototype consistency learning with consensus-based KAN fusion to reduce supervision bias.

It improves segmentation performance on public pancreas datasets, especially when annotations are scarce.

The work is relevant because accurate pancreas segmentation supports earlier cancer diagnosis and handles strong shape variation.