Abstract:Medical image segmentation has been significantly advanced with the rapid development of deep learning (DL) techniques. Existing DL-based segmentation models are typically discriminative; i.e., they aim to learn a mapping from the input image to segmentation masks. However, these discriminative methods neglect the underlying data distribution and intrinsic class characteristics, suffering from unstable feature space. In this work, we propose to complement discriminative segmentation methods with the knowledge of underlying data distribution from generative models. To that end, we propose a novel hybrid diffusion framework for medical image segmentation, termed HiDiff, which can synergize the strengths of existing discriminative segmentation models and new generative diffusion models. HiDiff comprises two key components: discriminative segmentor and diffusion refiner. First, we utilize any conventional trained segmentation models as discriminative segmentor, which can provide a segmentation mask prior for diffusion refiner. Second, we propose a novel binary Bernoulli diffusion model (BBDM) as the diffusion refiner, which can effectively, efficiently, and interactively refine the segmentation mask by modeling the underlying data distribution. Third, we train the segmentor and BBDM in an alternate-collaborative manner to mutually boost each other. Extensive experimental results on abdomen organ, brain tumor, polyps, and retinal vessels segmentation datasets, covering four widely-used modalities, demonstrate the superior performance of HiDiff over existing medical segmentation algorithms, including the state-of-the-art transformer- and diffusion-based ones. In addition, HiDiff excels at segmenting small objects and generalizing to new datasets. Source codes are made available at <a class="link-external link-https" href="https://github.com/takimailto/HiDiff" rel="external noopener nofollow">this https URL</a>.

Explicit-implicit priori knowledge-based diffusion model for generative medical image segmentation

MedSegDiff: Medical Image Segmentation with Diffusion Probabilistic Model

Edge-and-Mask Integration-Driven Diffusion Models for Medical Image Segmentation

Denoising Diffusions in Latent Space for Medical Image Segmentation

Accelerating Diffusion Models Via Pre-segmentation Diffusion Sampling for Medical Image Segmentation

BerDiff: Conditional Bernoulli Diffusion Model for Medical Image Segmentation

Cold SegDiffusion: A Novel Diffusion Model for Medical Image Segmentation

Diffusion Models for Implicit Image Segmentation Ensembles

PDDM: Poisson-Distribution Diffusion Model with Multi-Step Weighted Fusion for Medical Image Segmentation

PGDiffSeg: Prior-Guided Denoising Diffusion Model with Parameter-Shared Attention for Breast Cancer Segmentation

Anatomically-Controllable Medical Image Generation with Segmentation-Guided Diffusion Models

HiDiff: Hybrid Diffusion Framework for Medical Image Segmentation

Accelerating denoising diffusion probabilistic model via truncated inverse processes for medical image segmentation

DiffuseExpand: Expanding dataset for 2D medical image segmentation using diffusion models

Diffusion model-based text-guided enhancement network for medical image segmentation

Ambiguous Medical Image Segmentation using Diffusion Models

DiffSeg: A Segmentation Model for Skin Lesions Based on Diffusion Difference

FDiff-Fusion:Denoising diffusion fusion network based on fuzzy learning for 3D medical image segmentation

Distribution Aligned Diffusion and Prototype-guided network for Unsupervised Domain Adaptive Segmentation

Analysing Diffusion Segmentation for Medical Images

Diff-UNet: A Diffusion Embedded Network for Volumetric Segmentation