Date: 8 May 2026 (Friday)
Time: 11:00 am – 12:00 pm
Venue: Lecture Theatre 3/4, William M.W. Mong Block, 21 Sassoon Road
Host: Professor Michael Hӓusser

Zhihao Zheng studies connectomics—the comprehensive reconstruction of neural circuits at synaptic resolution using 3D electron microscopy (EM). During his PhD at HHMI Janelia research campus, he acquired the first EM volume of an entire adult fly brain, containing 50 million synapses and 139,255 neurons, followed by analysis of the olfactory circuits in the fly. As a postdoc jointly supervised by Drs. David Tank and Sebastian Seung at Princeton, he built a connectomic imaging facility with four fastest transmission electron microscopes. He has used these microscopes to generate the first connectomic dataset from the hippocampus, revealing a spatial gradient in excitatory connectivity and cell type–selective inhibitory connectivity in the mouse hippocampus CA3. He has co-led subsequent projects to acquire a 10x larger dataset of medial entorhinal cortex (~ 1 mm³) and to integrate connectomics with physiology, spatial transcriptomics, and whole-brain axonal projections.

To gain mechanistic insights into brain functions, I seek to map neural connectivity at a resolution sufficient to resolve individual synapses using 3D electron microscopy (EM connectomics). The hippocampus has been implicated in memory. While connectomics has achieved great successes with fly brains and mammalian cortices, previous EM volumes of hippocampus are either small or only sparsely reconstructed. We acquired a 3D EM volume (~1 × 1 × 0.1 mm³) and used artificial intelligence (AI)–based approaches to reconstruct the first connectomic dataset from hippocampal CA3. Within the dataset, we classified all neurons—including 1,815 pyramidal cells and 229 inhibitory cells. The analysis revealed a spatial gradient in excitatory connectivity and cell type–selective inhibitory connectivity that underlie the physiology and function of hippocampal CA3. These findings are related to properties in physiology, spatial transcriptomics, and whole-brain axonal projections. The dataset is shared on a public platform, pyr.ai.

All are welcome.