# Atlases Are Already Inside: Recovering Population Templates from Pretrained Diffusion Models

arXiv: https://arxiv.org/abs/2609.30566 · HTML: https://mlsanity.com/paper/2609.30566/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Medical imaging and segmentation, one of 14 directions in Diffusion models.
- Builds on 24 of its 47 references that are in the map.
- First posted on arXiv 24 Sep 2026.

## Placed in

- Diffusion models / [Medical imaging and segmentation](https://mlsanity.com/diffusion/medical-imaging-and-segmentation/)

## Builds on (most cited first, 24 of 24)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [Denoising Diffusion Implicit Models](https://mlsanity.com/paper/2010.02502/) (2020)
- [Score-Based Generative Modeling through Stochastic Differential Equations](https://mlsanity.com/paper/2011.13456/) (2020)
- [Deep Learning Face Attributes in the Wild](https://mlsanity.com/paper/1411.7766/) (2014)
- [SDEdit: Guided Image Synthesis and Editing with Stochastic Differential Equations](https://mlsanity.com/paper/2108.01073/) (2021)
- [VoxelMorph: A Learning Framework for Deformable Medical Image Registration](https://mlsanity.com/paper/1809.05231/) (2018)
- [Emergent Correspondence from Image Diffusion](https://mlsanity.com/paper/2306.03881/) (2023)
- [An Edit Friendly DDPM Noise Space: Inversion and Manipulations](https://mlsanity.com/paper/2304.06140/) (2023)
- [Diffusion Hyperfeatures: Searching Through Time and Space for Semantic Correspondence](https://mlsanity.com/paper/2305.14334/) (2023)
- [Learn2Reg: comprehensive multi-task medical image registration challenge, dataset and evaluation in the era of deep learning](https://mlsanity.com/paper/2112.04489/) (2021)
- [Deformed Implicit Field: Modeling 3D Shapes with Learned Dense Correspondence](https://mlsanity.com/paper/2011.13650/) (2020)
- [Deep Implicit Templates for 3D Shape Representation](https://mlsanity.com/paper/2011.14565/) (2020)
- [DeepAtlas: Joint Semi-Supervised Learning of Image Registration and Segmentation](https://mlsanity.com/paper/1904.08465/) (2019)
- [Learning Conditional Deformable Templates with Convolutional Networks](https://mlsanity.com/paper/1908.02738/) (2019)
- [GAN-Supervised Dense Visual Alignment](https://mlsanity.com/paper/2112.05143/) (2021)
- [DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model](https://mlsanity.com/paper/2112.05149/) (2021)
- [Medical Diffusion: Denoising Diffusion Probabilistic Models for 3D Medical Image Generation](https://mlsanity.com/paper/2211.03364/) (2022)
- [KeypointNet: A Large-scale 3D Keypoint Dataset Aggregated from Numerous Human Annotations](https://mlsanity.com/paper/2002.12687/) (2020)
- [Generative Adversarial Registration for Improved Conditional Deformable Templates](https://mlsanity.com/paper/2105.04349/) (2021)
- [Neural Congealing: Aligning Images to a Joint Semantic Atlas](https://mlsanity.com/paper/2302.03956/) (2023)
- [Aladdin: Joint Atlas Building and Diffeomorphic Registration Learning with Pairwise Alignment](https://mlsanity.com/paper/2202.03563/) (2022)
- [MultiMorph: On-demand Atlas Construction](https://mlsanity.com/paper/2504.00247/) (2025)
- [Continual atlas-based segmentation of prostate MRI](https://mlsanity.com/paper/2311.00548/) (2023)
- [AtlasMorph: Learning conditional deformable templates for brain MRI](https://mlsanity.com/paper/2511.13609/) (2025)

## Abstract (arXiv)

We present a new inference-time sampler for diffusion models that gives a pretrained model a capability it was never trained for: constructing the atlas of the population it synthesizes. The sampler converges from every random seed to the population's central anatomy, which we call the intrinsic atlas. The advantage is threefold. (1) It requires no retraining. A diffusion model that has already learned a coherent population, including the released ones, yields its atlas in a single inference pass without involving deformable registration. (2) It applies to multiple domains, such as brain MRI, chest X-ray, faces, and 3D shapes. (3) It extends to subpopulations. One age-conditioned model gives an atlas at any age in its training range, and the resulting family reproduces the CSF expansion of healthy aging. Evaluated as a registration target, the intrinsic atlas is best or second-best on every dataset against classical and learned templates, and the most central template on held-out brain MRI cohorts. Atlas construction can be reframed as a byproduct of generative modeling: a diffusion model is a learned representation of population structure, and the atlas is what it already contains.