Rosalind Franklin Institute, UK

报告摘要：

Recent advances in electron microscopy have greatly enhanced our ability to acquire high-dimensional datasets, often reaching several terabytes per hour. These data-rich experiments offer unprecedented opportunities to uncover critical material properties at the atomic scale, including three-dimensional atomic arrangements, elemental distributions, and local electronic states. However, the quantum-mechanical nature of electron–specimen interactions leads to highly entangled signals, while experimental artefacts introduced during acquisition further obscure the underlying structural information. As a result, accurately reconstructing specimen properties remains a major challenge.

In this presentation, I will explore how a combination of computational techniques can be used to maximise information extraction from complex electron microscopy datasets. I will begin by introducing the highly efficient open-source C++/CUDA software MULTEM, which leverages GPU acceleration to simulate electron–specimen interactions with high precision.

I will then demonstrate how machine learning can be integrated with tomography to recover atomic structures from noisy or incomplete data. In addition, I will show how image restoration techniques significantly improve the quality of ptychographic reconstructions. Finally, I will present validation strategies for machine learning–based restoration methods using realistically simulated multislice datasets informed by molecular dynamics.

Together, these approaches form a powerful and integrated computational framework for enhancing the accuracy, interpretability, and information retrieval capabilities of modern electron microscopy.

报告人简介：

Dr Iván Lobato is a physicist originally trained at the University of Antwerp in Belgium. He is currently based at the Rosalind Franklin Institute in Oxford, where his research focuses on the development of advanced computational methods and algorithms for quantitative electron microscopy. His work includes the application of deep learning for image restoration and analysis, the development of algorithms for 3D tomography and ptychography, and the creation of high-performance simulation software for modelling electron–specimen interactions.

邀请人：陈震 特聘研究员（82649878）

联系人：陈震 zhen.chen@iphy.ac.cn 010-82649878