Improving Multislice Electron Ptychography with a Generative Prior

TL;DR

This paper introduces MEP-Diffusion, a diffusion model that acts as a generative prior to improve the quality and efficiency of multislice electron ptychography reconstructions, significantly outperforming existing methods.

Contribution

The paper presents MEP-Diffusion, a novel diffusion-based prior trained on crystal structures, integrated into reconstruction algorithms via Diffusion Posterior Sampling to enhance 3D volume quality.

Findings

Achieved a 90.50% improvement in SSIM over existing methods.

Enhanced reconstruction quality and efficiency in multislice electron ptychography.

Successfully integrated the diffusion model as a prior into existing algorithms.

Abstract

Multislice electron ptychography (MEP) is an inverse imaging technique that computationally reconstructs the highest-resolution images of atomic crystal structures from diffraction patterns. Available algorithms often solve this inverse problem iteratively but are both time consuming and produce suboptimal solutions due to their ill-posed nature. We develop MEP-Diffusion, a diffusion model trained on a large database of crystal structures specifically for MEP to augment existing iterative solvers. MEP-Diffusion is easily integrated as a generative prior into existing reconstruction methods via Diffusion Posterior Sampling (DPS). We find that this hybrid approach greatly enhances the quality of the reconstructed 3D volumes, achieving a 90.50% improvement in SSIM over existing methods.