BIOMEDICAL IMAGING GROUP (BIG)
Laboratoire d'imagerie biomédicale (LIB)

Opportunities and Challenges for Generative Adversarial Reconstruction by Distribution Matching (CryoGAN)

M. Unser, P. Bohra

SIAM Conference on Imaging Science (IS'22), Virtual, March 21-25, 2022, MS96, Single-Particle Cryo-Electron Microscopy Recent Advances and Next Challenges - Part II of III.

Single-particle cryo-electron microscopy (cryo-EM) has revolutionised the field of structural biology over the last decade, culminating in 2017 by the awarding of the Nobel Prize in Chemistry to its three founders. Nowadays, single-particle cryo-EM permits the regular discovery of new biological structures at atomic resolution. Yet, the reconstruction task remains an enduring challenge due to the unknown orientations adopted by the 3D particles prior to imaging. CryoGAN is a recently proposed generative adversarial reconstruction paradigm that is based on distribution matching and allows us to sidestep pose estimation. This new paradigm presents us with an opportunity to reconstruct the entire continuum of 3D conformations of the protein, which is an outstanding problem in structural biology. However, currently CryoGAN is unable to match the resolution achieved by the state-of-the-art methods. In order to understand this scheme in depth and improve its performance, we study its application to the simpler problem of single particle image refinement. Our experiments indicate that CryoGAN is able to match statistical methods such as maximum likelihood estimation up to a certain amount of noise beyond which the reconstruction deteriorates. Thus, this presents us with the challenge of designing a more sophisticated version of the CryoGAN that is more robust to noise and which can then be potentially used for the 3D continuous conformation reconstruction problem. 

@INPROCEEDINGS(http://bigwww.epfl.ch/publications/unser2203.html,
AUTHOR="Unser, M. and Bohra, P.",
TITLE="Opportunities and Challenges for Generative Adversarial
	Reconstruction by Distribution Matching ({CryoGAN})",
BOOKTITLE="{SIAM} Conference on Imaging Science ({IS'22})",
YEAR="2022",
editor="",
volume="",
series="",
pages="",
address="Virtual",
month="March 21-25,",
organization="",
publisher="",
note="MS96")
© 2022 SIAM. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from SIAM. This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.