Research map / 3D vision / Face and shape
CryoAI: Amortized Inference of Poses for Ab Initio Reconstruction of 3D Molecular Volumes from Real Cryo-EM Images
Axel Levy, Frédéric Poitevin, Julien Martel, Youssef Nashed, Ariana Peck, Nina Miolane, Daniel Ratner, Mike Dunne, Gordon Wetzstein
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use amortized pose encoder instead of per-image pose search (cryoAI vs cryoSPARC) | cryoAI · Simulated 80S, increasing dataset size | convergence time: cryoSPARC reaches 5 hours with 9M particles; cryoAI time independent of dataset size | Improved |
| add symmetric loss instead of L2 loss (ablation) | cryoAI · Noisy adenylate kinase (L=128) | volume reconstruction quality | Improved |
| add symmetric loss to cryoAI and cryoPoseNet (ablation) | cryoAI / cryoPoseNet · Noise-free adenylate kinase (L=64) | convergence time | Improved |
| operate in Fourier space (cryoAI) instead of real-space volume (cryoPoseNet) with the same loss | cryoAI · Noise-free adenylate kinase (L=64) | convergence time | Improved |
| use FourierNet instead of SIREN or ReLU MLP with positional encoding (ablation) | cryoAI decoder · Noisy adenylate kinase (L=128) | convergence speed | Improved |
| cryoAI vs cryoPoseNet | cryoAI · Spike (ideal) | resolution (pixels): 16.0 -> 2.29 | Improved |
| cryoAI vs cryoPoseNet | cryoAI · Spliceosome (noisy) | resolution (pixels): 3.15 -> 2.61 | Improved |
| cryoAI vs cryoSPARC with noise | cryoAI · Spike (noisy) | resolution (pixels): 3.56 -> 2.91 | Improved |
| cryoAI vs cryoDRGN2 with noise | cryoAI · Spike (noisy) | resolution (pixels): 2.03 -> 2.91 | Hurt |
| add noise to spliceosome data | cryoAI · Spliceosome | resolution (pixels): 2.13 -> 2.61 | Hurt |
| add noise to spike data | cryoAI · Spike | resolution (pixels): 2.29 -> 2.91 | Hurt |
| cryoAI vs cryoSPARC on rotation error | cryoAI · Spike (noisy) | rotation error: 0.02 -> 0.01 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- cryoDRGN2 — CryoDRGN2: Ab Initio Neural Reconstruction of 3D Protein Structures From Real Cryo-EM Images (ICCV 2021) · better on 0 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- CryoSPIN: Improving Ab-Initio Cryo-EM Reconstruction with Semi-Amortized Pose Inference (NeurIPS 2024 · similarity 0.94)
- cryoSPHERE: Single-Particle HEterogeneous REconstruction from cryo EM (ICLR 2025 · similarity 0.92)
- CryoBench: Diverse and challenging datasets for the heterogeneity problem in cryo-EM (NeurIPS 2024 · similarity 0.92)
- Mixture of neural fields for heterogeneous reconstruction in cryo-EM (NeurIPS 2024 · similarity 0.91)
- Multiscale guidance of protein structure prediction with heterogeneous cryo-EM data (NeurIPS 2025 · similarity 0.91)
- Reconstructing Heterogeneous Biomolecules via Hierarchical Gaussian Mixtures and Part Discovery (NeurIPS 2025 · similarity 0.91)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
