Coming soon: Workshop on Computational Microscopy at IEEE ICIP 2026 (13–17 Sept, Tampere, Finland). Details

Research

We develop theory, algorithms and microscopy systems that couple X-ray and electron imaging with AI for advanced materials characterization.

Structured X-ray illumination for nanoscale chemical imaging (NxSCI)
Figure from NxSCI (2023)

Computational X-ray Microscopy

We build AI-driven pipelines for X-ray ptychography, tomography, spectro-tomography, and laminography. The goal is quantitative structure and chemistry from incomplete, noisy, or probe-unknown measurements — from synchrotron beamlines to in-device experiments.

Selected papers

ENCODE continuous neural representations for electron microscopy
Figure from ENCODE (2026)

Computational Electron Microscopy

We also push the imaging and analysis limits of electron microscopy. Continuous neural representations and learning-based reconstruction turn large, noisy, or incomplete EM measurements into compact, high-fidelity structure — from scalable data representation to quantitative microstructure recovery.

Selected papers

Multi-scale cathode particle network and damage progression from X-ray imaging
Figure from Li, Science (2022)

AI for Battery Science

We use computational microscopy for distinctive battery characterization, revealing otherwise invisible structure and dynamics—especially during operando experiments—for failure analysis, cathode microstructure, particle networks, and explainable models of thermochemistry and degradation.

Selected papers

High-frequency voltage imaging of neuronal dynamics (TEMPO)
Figure from Haziza et al., Cell (2025)

Inverse Problem for Imaging

Across modalities, we treat imaging as an inverse problem: denoising, phase retrieval, probe recovery, sparse-view tomography, and neural representations. Methods developed here transfer between X-ray, electron, optical microscopy and related imaging systems.

Selected papers