KANResDiff: Learning Local Residual Diffusion via Kolmogorov-Arnold Network for Ambiguous Medical Image Segmentation

2026-08-11

AI Summary

This work addresses the limitations of existing fuzzy medical image segmentation methods, which introduce randomness in a fixed manner and lack progressive semantic modeling. To overcome these issues, the authors propose a stage-aware diffusion framework that leverages Kolmogorov–Arnold networks to learn a local residual diffusion process. The approach employs spline-based time embeddings to enable independent temporal encoding, thereby enhancing semantic distinctions across stages. Furthermore, a learnable-weight residual Schrödinger bridge is introduced to inject deterministic priors, facilitating the construction of locally optimal diffusion trajectories. Evaluated on two public datasets, the model achieves state-of-the-art performance, improving the GED and HM-IoU metrics by 16.8% and 7.7%, respectively, while maintaining competitive results on the MDM metric.

Abstract

Ambiguous medical image segmentation aims to provide a series of diverse but plausible segmentation hypotheses. However, existing methods introduce stochasticity in a fixed and pre-defined manner, failing to form a progressive semantic modeling process. To address these challenges, we propose KANResDiff to learn local residual diffusion with Kolmogorov-Arnold Network, thereby assigning distinct roles across stages for ambiguity modeling. Specifically, we propose Independent Time Encoding that offers spline-based time embeddings instead of linear ones from MLPs, which enhances the independence across inference stages and assigns progressive semantic roles to different stages. We propose Residual Schrodinger Bridge that injects deterministic residual prior with learnable weights by constructing local Schrodinger Bridge instead of following manually settings, achieving a flexible deterministic-stochastic interaction and stage-aware ambiguity modeling thanks to local optimal diffusion path. Extensive experimental results on two public datasets demonstrate that KANResDiff achieves SOTA performance on GED and HM-IoU, with maximum improvements of 16.8% and 7.7%, respectively, while maintaining competitive performance on the MDM metric. Source code is available at https://github.com/PerceptionComputingLab/KANResDiff.

Problem

• ambiguous medical image segmentation
• stochasticity
• semantic modeling
• segmentation hypotheses
• diffusion

Innovation

• Kolmogorov-Arnold Network
• Local Residual Diffusion
• Independent Time Encoding
• Residual Schrodinger Bridge
• Ambiguous Medical Image Segmentation