About Open Cryo-ET

The physics of imaging, and the statistical machine learning behind reconstruction — in one place, made clear.
This is not another tool's documentation. It is an encyclopedia of why.

Cryo-ET lets us see inside a cell close to its native state — but its 3-D reconstructions are hard to read: a ±60° tilt leaves a missing wedge that stretches structure along one axis, and the low dose that protects the sample buries the signal in noise. The missing wedge is not noise; it is a whole block of information that was never measured. Filling it back in takes both the physics of imaging and probabilistic, generative modelling — and this site develops both sides at once.

How to read it

We want two kinds of reader to get through it: the one who knows only biology, and the one who knows only machine learning. Every concept opens with an Intuition that builds the picture, then a Depth section with the derivation; many pages carry a demo computed live in your browser. You need not absorb all the math at once — follow the nine knowledge bases down, and step back to the prerequisites whenever a page loses you. The further you read, the clearer the missing wedge becomes.

Our work

The site also develops our own family of self-supervised Cryo-ET reconstruction methods — no ground-truth labels, missing-wedge restoration written as a Bayesian inverse problem. CryoGEN-I is a MAP point estimate under an energy prior[1]; CryoGEN-II turns to optimal transport for global distribution matching[2]; CryoWGEN adds entropic regularization for an uncertainty-capturing Boltzmann posterior. They focus on reconstruction itself: turning that missing block back into a more isotropic, more complete volume.

How to cite

If this site or these methods help your work, please cite:

1. Yunfei Teng, Yuxuan Ren, Kai Chen, Xi Chen, Zhaoming Chen, Qiwei Ye. CryoGEN: Generative Energy-based Models for Cryogenic Electron Tomography Reconstruction. ICLR, 2025. openreview.net/forum?id=uOb7rij7sR
2. Yunfei Teng et al. CryoGEN-II: Cryogenic Electron Tomography Reconstruction via Generative Network. CVPR 2026 Workshops (VISION). openaccess.thecvf.com

@inproceedings{teng2025cryogen,
  title     = {CryoGEN: Generative Energy-based Models for
               Cryogenic Electron Tomography Reconstruction},
  author    = {Teng, Yunfei and Ren, Yuxuan and Chen, Kai and
               Chen, Xi and Chen, Zhaoming and Ye, Qiwei},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year      = {2025},
  url       = {https://openreview.net/forum?id=uOb7rij7sR}
}

@inproceedings{teng2026cryogen2,
  title     = {CryoGEN-II: Cryogenic Electron Tomography
               Reconstruction via Generative Network},
  author    = {Teng, Yunfei and others},
  booktitle = {CVPR 2026 Workshops (VISION)},
  year      = {2026},
  url       = {https://openaccess.thecvf.com/content/CVPR2026W/VISION26/papers/Teng_CryoGEN-II_Cryogenic_Electron_Tomography_Reconstruction_via_Generative_Network_CVPRW_2026_paper.pdf}
}

Team & contact

Open Cryo-ET is maintained by the Open-Cryo Team — an open-source academic team that works on Cryo-ET reconstruction methods and builds this site to explain the technique alongside the statistical machine learning behind it. We believe good science should be open, readable, and within everyone's reach: the site is fully bilingual (switch in the top right) and every interactive demo is free to use. Currently in trial run, still being polished. Corrections, additions, and collaborations are welcome — yt1208@nyu.edu.