Anatomica: Localized Control over Geometric and Topological Properties for Anatomical Diffusion Models

TL;DR

Anatomica introduces a flexible inference-time framework that enables precise control over geometric and topological features of anatomical diffusion models using localized constraints and persistent homology.

Contribution

It is the first framework to provide localized, multi-class control over anatomical structures during diffusion model generation using differentiable penalties.

Findings

Effective control of size, shape, and position of structures.

Enables enforcement of topological features like loops and voids.

Applicable across diverse anatomical systems.

Abstract

We present Anatomica: an inference-time framework for generating multi-class anatomical voxel maps with localized geo-topological control. During generation, we use cuboidal control domains of varying dimensionality, location, and shape to slice out relevant substructures. These local substructures are used to compute differentiable penalty functions that steer the sample towards target constraints. We control geometric features such as size, shape, and position through voxel-wise moments, while topological features such as connected components, loops, and voids are enforced through persistent homology. Lastly, we implement Anatomica for latent diffusion models, where neural field decoders partially extract substructures, enabling the efficient control of anatomical properties. Anatomica applies flexibly across diverse anatomical systems, composing constraints to control complex structures over arbitrary dimensions and coordinate systems, thereby enabling the rational design of synthetic datasets for virtual trials or machine learning workflows.