Electron Microscopy Tomography

Presented by

Presented by Albert Lawrence, James Bouwer, University of California, San Diego

Abstract

In recent years the biological electron microscopy community has adapted technical improvements in instrumentation, data collection modes and large format digital image detectors. For example images on the order of 8K X 8K pixels are becoming commonplace, and multiple detectors and montaging techniques make series of multi-billion pixel images achievable without extraordinary effort. As input to tomographic reconstruction this flood of image data plays an important role in determining the three-dimensional structure and function of cells and sub-cellular organelles. Structure may be elucidated across a wide range of spatial scales, ranging from that of neurons in the brain, for example, down to the scale of proteins and protein complexes. The scale and volume of this data, is in itself, a challenge to the transmission, reconstruction and storage of the processed data. In response to the requirements for high-quality three-dimensional reconstructions from EM data this branch of computer tomography is presently under a state of rapid development. Areas of future progress include instrumentation, data collection, reconstruction and other image processing techniques.

Electron microscope (EM) tomography presents a number of special problems. The imagery is low contrast and noisy with limited sampling of projection directions; sample warping and the curvilinearity of electron trajectories make classical techniques of x-ray tomography problematic; and the volume and scale of the data make automated preprocessing and image segmentation necessary. The need for solution of these problems has spurred the introduction of new techniques into EM tomography. For example, the presence of geometric nonlinearities in the basic ray transform requires that the inversion problem be treated in terms of Fourier integral operators rather than Fourier transforms. Recently Bayesian techniques and Markov random field techniques have been applied to other steps in the tomographic reconstruction process such as feature extraction, tracking, alignment, and post processing steps. Progress in all aspects of EM tomography requires deeper mathematical understanding, and the introduction of new algorithms and techniques from computer science.

Although we will emphasize applications of EM tomography to the biological sciences, materials science also makes extensive use of this technology. Electron tomography in the materials sciences is a rapidly advancing area of scientific endeavor and affords many opportunities for collaborative activities.

This tutorial will focus on new developments in Electron Microscope instrumentation, applications to the biological sciences, and challenges presented by recent developments in obtaining high-quality large field reconstructions.

Topics will include:

Speaker Biographies

James Bouwer received his PhD in Physics from the University of California, San Diego in 2001. Currently James is a principal development engineer at the National Center for Microscopy and Imaging Research at UCSD. He spends most of his time working on microscope automation and various problems in electron microscopy and tomography. He has recently completed the development and implementation of the worlds highest resolution 8k x 8k lens coupled camera system for TEM.

Albert Lawrence received his PhD in Mathematics from the University of Chicago in 1969. He has been hanging around biology labs for most of the past 40 years. Highlights of his career include new statistical methods EEG signal processing, electronic circuit modeling, application of photo-active molecules to molecular electronics, and development of an advanced code for electron microscope tomography.