Prior-Guided Residual Diffusion: Calibrated and Efficient Medical Image Segmentation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mao, Fuyou, Wu, Beining, Jiang, Yanfeng, Xue, Han, Tang, Yan, Zhang, Hao
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914297125273600
author Mao, Fuyou
Wu, Beining
Jiang, Yanfeng
Xue, Han
Tang, Yan
Zhang, Hao
author_facet Mao, Fuyou
Wu, Beining
Jiang, Yanfeng
Xue, Han
Tang, Yan
Zhang, Hao
contents Ambiguity in medical image segmentation calls for models that capture full conditional distributions rather than a single point estimate. We present Prior-Guided Residual Diffusion (PGRD), a diffusion-based framework that learns voxel-wise distributions while maintaining strong calibration and practical sampling efficiency. PGRD embeds discrete labels as one-hot targets in a continuous space to align segmentation with diffusion modeling. A coarse prior predictor provides step-wise guidance; the diffusion network then learns the residual to the prior, accelerating convergence and improving calibration. A deep diffusion supervision scheme further stabilizes training by supervising intermediate time steps. Evaluated on representative MRI and CT datasets, PGRD achieves higher Dice scores and lower NLL/ECE values than Bayesian, ensemble, Probabilistic U-Net, and vanilla diffusion baselines, while requiring fewer sampling steps to reach strong performance.
format Preprint
id arxiv_https___arxiv_org_abs_2509_01330
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Prior-Guided Residual Diffusion: Calibrated and Efficient Medical Image Segmentation
Mao, Fuyou
Wu, Beining
Jiang, Yanfeng
Xue, Han
Tang, Yan
Zhang, Hao
Computer Vision and Pattern Recognition
Ambiguity in medical image segmentation calls for models that capture full conditional distributions rather than a single point estimate. We present Prior-Guided Residual Diffusion (PGRD), a diffusion-based framework that learns voxel-wise distributions while maintaining strong calibration and practical sampling efficiency. PGRD embeds discrete labels as one-hot targets in a continuous space to align segmentation with diffusion modeling. A coarse prior predictor provides step-wise guidance; the diffusion network then learns the residual to the prior, accelerating convergence and improving calibration. A deep diffusion supervision scheme further stabilizes training by supervising intermediate time steps. Evaluated on representative MRI and CT datasets, PGRD achieves higher Dice scores and lower NLL/ECE values than Bayesian, ensemble, Probabilistic U-Net, and vanilla diffusion baselines, while requiring fewer sampling steps to reach strong performance.
title Prior-Guided Residual Diffusion: Calibrated and Efficient Medical Image Segmentation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2509.01330