Beyond Euclidean Prototypes: Spectral Disentanglement and Geodesic Matching for Few-Shot Medical Image Segmentation

2026-05-18

AI Summary

This work addresses the challenges of fragmented segmentation and tissue leakage in few-shot medical image segmentation, which arise from prototype cue coupling and the neglect of feature manifold structure in Euclidean-space matching. To overcome these limitations, we propose SGP-Net, which introduces learnable radial Fourier filters to decouple shape, texture, and boundary prototypes in the spectral domain. A feature affinity graph is constructed, and differentiable graph propagation matching is achieved via heat diffusion to approximate geodesic distances, replacing conventional cosine similarity. This design effectively preserves both semantic consistency and manifold topology. Extensive experiments on three public benchmarks demonstrate that SGP-Net achieves state-of-the-art performance, significantly alleviating response fragmentation within low-contrast organs and leakage into adjacent tissues.

Abstract

Few-Shot Medical Image Segmentation (FSMIS) aims to delineate novel anatomical targets from one or a few annotated support images, addressing the annotation scarcity in medical imaging. Notwithstanding recent advancements, current prototype-based methods are bottlenecked by two coupled limitations: 1) cue entanglement, where a single spatial-domain prototype is forced to summarise organ silhouette, parenchymal texture and boundary appearance simultaneously, so any support-query mismatch on one cue propagates indiscriminately to the others; and 2) topology-blind matching, where cosine similarity measures distance in the ambient Euclidean space and ignores the connectivity of the underlying feature manifold, causing fragmented activations inside low-contrast organs and leakage into neighbouring tissues. To this end, we propose Spectral-Geodesic Prototype Network (SGP-Net), built around a Spectral-Geodesic Prototype Module with two coupled components. A Spectral Prototype Bank (SPB) decomposes support and query features into low-, mid- and high-frequency bands via learnable radial Fourier filters, yielding three disentangled prototypes per class that separately encode shape, texture and boundary cues. A Geodesic Matcher (GM) then replaces cosine similarity with a differentiable heat-diffusion approximation of geodesic distance, propagating matching signals along a feature affinity graph so that on-manifold pixels accumulate consistent responses while off-manifold look-alikes are suppressed. Experiments on three public FSMIS benchmarks demonstrate that SGP-Net achieves competitive performance against recent state-of-the-art methods.

Problem

• cue entanglement
• topology-blind matching
• few-shot medical image segmentation
• prototype-based methods
• feature manifold

Innovation

• spectral disentanglement
• geodesic matching
• few-shot segmentation
• prototype network
• manifold-aware similarity