Assistant Professor Lin Gu participated in the IEEE International Symposium on Biomedical Imaging (held in London) and presented the latest research findings on AI for Science. This conference is one of the leading international conferences in the field of bioimaging, and many cutting-edge research projects are reported there. In this presentation, he introduced AI technology for drug discovery support utilizing electron microscope data.
The presentation covered the following two topics:
- Toward Efficient End-to-End VEM Processing Using a Unified Agent on GPUs and NPUs
He proposed an AI agent that comprehensively manages a series of processes in electron microscopy analysis (file management, stitching, registration, inpainting, slice reconstruction, etc.). Optimized for parallel processing on a GPU cloud environment, it achieves up to 10 times faster processing compared to traditional manual workflows. - Detection over Segmentation: A New Approach for Multi-Particle Picking in 3D Real-World CryoET
He proposed a new AI method for automatically detecting particles with high accuracy in cryo-electron tomography (cryo-ET) data in real-world environments. This new approach offers a balance of efficiency and accuracy, replacing conventional segmentation-based methods.
Furthermore, He gave an invited lecture at “Exploring Foundation Models in Medical Image Analysis: Applications, Challenges, and Uncertainties,” where he introduced the challenges of foundation models in the medical field and my unique approach to solving them. He also reported on the latest advances in AI analysis of brain imaging such as fMRI.
