Volumetric Directional Diffusion: Anchoring Uncertainty Quantification in Anatomical Consensus for Ambiguous Medical Image Segmentation

TL;DR

This paper introduces Volumetric Directional Diffusion (VDD), a novel generative model for 3D medical image segmentation that effectively captures uncertainty and anatomical variability while maintaining structural integrity.

Contribution

VDD uniquely anchors the generative process to a deterministic consensus prior, improving uncertainty quantification and anatomical fidelity in ambiguous medical image segmentation.

Findings

VDD outperforms existing methods in uncertainty metrics (GED, CI).

VDD maintains high segmentation accuracy comparable to deterministic models.

VDD provides anatomically coherent uncertainty maps for clinical decision-making.

Abstract

Equivocal 3D lesion segmentation exhibits high inter-observer variability. Conventional deterministic models ignore this aleatoric uncertainty, producing over-confident masks that obscure clinical risks. Conversely, while generative methods (e.g., standard diffusion) capture sample diversity, recovering complex topology from pure noise frequently leads to severe structural fractures and out-of-distribution anatomical hallucinations. To resolve this fidelity-diversity trade-off, we propose Volumetric Directional Diffusion (VDD). Unlike standard diffusion models that denoise isotropic Gaussian noise, VDD mathematically anchors the generative trajectory to a deterministic consensus prior. By restricting the generative search space to iteratively predict a 3D boundary residual field, VDD accurately explores the fine-grained geometric variations inherent in expert disagreements without risking topological collapse. Extensive validation on three multi-rater datasets (LIDC-IDRI, KiTS21, and ISBI 2015) demonstrates that VDD achieves state-of-the-art uncertainty quantification (significantly improving GED and CI) while remaining highly competitive in segmentation accuracy against deterministic upper bounds. Ultimately, VDD provides clinicians with anatomically coherent uncertainty maps, enabling safer decision-making and mitigating risks in downstream tasks (e.g., radiotherapy planning or surgical margin assessment).