Autonomous Electron Tomography Reconstruction with Machine Learning

TL;DR

This paper introduces an automated, efficient approach for electron tomography reconstruction that combines Bayesian optimization with momentum-enhanced compressed sensing, significantly reducing computation time and improving reconstruction quality.

Contribution

It presents a novel method integrating Bayesian optimization and momentum-based gradient descent to automate parameter selection and enhance reconstruction quality in electron tomography.

Findings

80% reduction in compute time for 3D nanocube reconstruction

High-quality tomograms achieved with heavily weighted TV minimization

Over-smoothing avoided by adding momentum to gradient descent

Abstract

Modern electron tomography has progressed to higher resolution at lower doses by leveraging compressed sensing methods that minimize total variation (TV). However, these sparsity-emphasized reconstruction algorithms introduce tunable parameters that greatly influence the reconstruction quality. Here, Pareto front analysis shows that high-quality tomograms are reproducibly achieved when TV minimization is heavily weighted. However, in excess, compressed sensing tomography creates overly smoothed 3D reconstructions. Adding momentum to the gradient descent during reconstruction reduces the risk of over-smoothing and better ensures that compressed sensing is well behaved. For simulated data, the tedious process of tomography parameter selection is efficiently solved using Bayesian optimization with Gaussian processes. In combination, Bayesian optimization with momentum-based compressed sensing greatly reduces the required compute time$-$an 80% reduction was observed for the 3D reconstruction of SrTiO$_3$ nanocubes. Automated parameter selection is necessary for large scale tomographic simulations that enable the 3D characterization of a wider range of inorganic and biological materials.