Ruyi Lian, Ph.D. Research Proficiency Presentation: 'End-to-end orientation estimation from 2D cryo-EM images'

Wednesday, July 21, 2021 - 11:30am to 1:00pm
Zoom - contact for Zoom info.
Event Description: 


Cryo-electron microscopy (cryo-EM) is a Nobel Prize-winning technique for determining high-

resolution 3D structures of biological macromolecules. A 3D structure is reconstructed from a

tremendous number of noisy 2D projection images. However, existing 3D reconstruction methods are still time-consuming. One of the major computational bottlenecks is to recover the unknown particle pose from each 2D image, which is usually determined by expensive global search.

In this talk, we will first review the existing pose estimation methods from RGB images and cryo-EM images. Inspired by the development of pose estimation from RGB images, we then present a novel end-to-end supervised learning method for estimating 3D orientations from 2D cryo-EM images. We utilize a neural network to approximate the mapping from images to orientations, and propose a robust loss function to handle both asymmetric and symmetric 3D structures. Experiments on synthetic datasets with various symmetry types confirm that the neural network is capable of recovering orientations from 2D images, and the results on one real cryo-EM dataset further demonstrate its potential in more challenging imaging conditions.

Computed Event Type: