3D Gaussian Splatting for Annular Dark Field Scanning Transmission Electron Microscopy Tomography Reconstruction

2026-04-06

AI Summary

This study addresses the artifacts and structural distortions in annular dark-field scanning transmission electron microscopy (ADF-STEM) tomography caused by sparse angular sampling. To this end, it introduces 3D Gaussian Splatting (3DGS) into the field for the first time, proposing a learnable local scattering intensity scalar field (denza), a scattering-angle-dependent normalization coefficient γ, and a loss function incorporating 2D Fourier amplitude information. The method effectively suppresses missing-wedge artifacts and achieves high-fidelity 3D reconstructions using only 15–45 tilt angles. The resulting projections exhibit strong consistency with the original experimental images, significantly outperforming conventional reconstruction approaches.

Abstract

Analytical Dark Field Scanning Transmission Electron Microscopy (ADF-STEM) tomography reconstructs nanoscale materials in 3D by integrating multi-view tilt-series images, enabling precise analysis of their structural and compositional features. Although integrating more tilt views improves 3D reconstruction, it requires extended electron exposure that risks damaging dose-sensitive materials and introduces drift and misalignment, making it difficult to balance reconstruction fidelity with sample preservation. In practice, sparse-view acquisition is frequently required, yet conventional ADF-STEM methods degrade under limited views, exhibiting artifacts and reduced structural fidelity. To resolve these issues, in this paper, we adapt 3D GS to this domain with three key components. We first model the local scattering strength as a learnable scalar field, denza, to address the mismatch between 3DGS and ADF-STEM imaging physics. Then we introduce a coefficient $γ$ to stabilize scattering across tilt angles, ensuring consistent denza via scattering view normalization. Finally, We incorporate a loss function that includes a 2D Fourier amplitude term to suppress missing wedge artifacts in sparse-view reconstruction. Experiments on 45-view and 15-view tilt series show that DenZa-Gaussian produces high-fidelity reconstructions and 2D projections that align more closely with original tilts, demonstrating superior robustness under sparse-view conditions.

Problem

• ADF-STEM
• tomography reconstruction
• sparse-view
• electron dose sensitivity
• missing wedge artifacts

Innovation

• 3D Gaussian Splatting
• ADF-STEM tomography
• sparse-view reconstruction
• scattering normalization
• missing wedge artifact suppression