A research center at China Medical University · Taichung, Taiwan  ·  Contact us →
Spatiomics · AI · Clinical Medicine

Making the brain
computable across scales.

We integrate multiscale 3D imaging, spatial omics, artificial intelligence, high-performance computing, and clinical medicine — building comprehensive brain maps that turn complex biological data into biomedical intelligence.

Explore Research → About the Center
Scroll
01

Multiscale 3D Imaging

From molecules to systems, we visualize the brain in three dimensions across multiple spatial scales.

02

Spatial Omics

We map genes, proteins, and cells in their native spatial context to reveal the language of the brain.

03

AI & High-Performance Computing

We develop AI models and scalable pipelines to transform massive imaging and omics data into insight.

04

Clinical Medicine

We connect brain maps with clinical knowledge to advance precision medicine and brain health.

Prof. Ann-Shyn Chiang
Founder & Director

Prof. Ann-Shyn Chiang

江 安 世 院士
President, China Medical University · Academician, Academia Sinica
Center Introduction

Transforming biological structure into digital intelligence.

The Brain Spatiomics Center (BSC) at China Medical University integrates multiscale three-dimensional imaging, spatial omics, artificial intelligence, high-performance computing, and clinical medicine to transform biological structure into digital intelligence.

BSC is designed not simply as an imaging core, but as a discovery and translation platform for understanding how the brain's physical organization gives rise to computation, memory, behavior, and neurological disease.

Spatiomics is the Center's organizing principle. Structural wiring, gene expression, protein localization, electrophysiology, clinical phenotype, and longitudinal disease trajectories are brought into a common spatial framework. This enables biological specimens to become standardized, computable models that can be explored by scientists, interpreted by clinicians, and reused by AI systems.

160+
Publications
37+
Top-Tier Journals
<10nm
Resolution
Organization

Six divisions, one integrated platform.

An integrated research architecture spanning international outreach, data and AI, ethics, imaging factory, functional applications, and clinical translation.

— DIVISION 01

Brain Spatiomics International Outreach

International Outreach Division

Positioning CMU as a hub for brain spatiomics in Asia and globally — fostering collaboration, exchange, and resource integration.

K
Prof. & Deputy Director
Fu-Jen Kao
— DIVISION 02

Digital Data & AI Applications

Digital Data & AI Division

Developing AI pipelines and high-throughput automation for neuroscience data at petabyte scale.

S
Prof. & Deputy Director
Chi-Tin Shih
— DIVISION 03

Brain Science & AI Ethics

Ethics & Governance Division

Integrating philosophy, humanities, and technology for AI medicine, neuroethics, and tech governance.

C
Prof. & Deputy Director
Szu-Ting Chen
— DIVISION 04

Biomedical Tech. R&D & Imaging Factory

Imaging Factory Division

High-throughput, high-resolution imaging for multiscale observation and connectomics research.

C
Head · Research Fellow
Wei-Kun Chang
— DIVISION 05

Brain Spatiomics Functional Applications

Functional Applications Division

Converting structural atlases into functional understanding — bridging basic science and clinical research.

F
Head · Research Fellow
Kuan-Lin Feng
— DIVISION 06

Clinical Translation

Clinical Translation Division

An integrated clinical, translational, and basic research team to unlock complex neurological conditions.

C
Physician & Deputy Director
Jui-Cheng Chen
Mission

Decoding how the brain computes.

01

Decode neural computation

Reveal how molecules, cells, and circuits generate perception, memory, decisions, and behavior in model animals.

02

Build navigable human brain atlases

Create AI-enabled, high-resolution maps that connect tissue architecture with diagnosis and treatment.

AI Bioimaging Foundry

An AI-orchestrated, closed-loop infrastructure.

The Center's operational engine is the AI Bioimaging Foundry — an AI-orchestrated, closed-loop infrastructure that moves from specimen to validated biological knowledge. Standardized sample preparation, large-volume optical imaging, AI super-resolution, Multiview Expansion Tomography (MExT), AI reconstruction, GPU computing, curated knowledge bases, and physician-led validation are integrated into one reproducible pipeline.

MExT supplies high-resolution, near-isotropic three-dimensional reference data; AI/HPC converts these data into aligned volumes, segmented cells and processes, candidate synaptic features, and reconstructed circuits. Standard protocols, quality-control checkpoints, and data provenance make the workflow scalable, comparable, and reusable.

01 Sample Preparation
02 Large-Volume Optical Imaging
03 AI Super-Resolution & MExT
04 GPU Reconstruction & Computing
05 Physician-Led Validation
Research Focus

Five interlocking programs define our agenda.

BSC develops a cross-scale framework that connects molecular identity and cellular morphology with circuit architecture, computation, disease, and precision medicine.

— PROGRAM 01

Human Brain Spatiomics

Build AI super-resolution atlases of the human brain, beginning with hippocampal and cortical circuits and progressing toward multi-centimeter and ultimately whole-brain mapping.

Learn More
— PROGRAM 02

Comparative Connectomics

Integrate Drosophila, mouse, and human datasets to identify conserved and divergent principles of circuit organization, neural computation, learning, memory, and individualized behavior.

Learn More
— PROGRAM 03

Disease Spatiomics

Map cellular vulnerability, circuit disruption, and molecular pathology in epilepsy, neurodegenerative disease, and brain tumors.

Learn More
— PROGRAM 04

AI & Data Infrastructure

Create AI-ready 3D reference datasets, scalable reconstruction pipelines, and interoperable knowledge platforms for terabyte- to petabyte-scale brain data.

Learn More
— PROGRAM 05

Precision Neuromedicine

Translate spatial brain maps into physician-in-the-loop tools for biomarker stratification, treatment planning, and patient-specific digital brain models.

Learn More
Golgi-X7-ExM: Human frontal cortex neural network
Figure 1 Depth-coded three-dimensional projection of Golgi-stained human frontal cortex after sevenfold expansion microscopy (Golgi–X7-ExM). Color represents relative imaging depth across the volume.
Research Highlights

Technology in Action: Human Frontal Cortex.

Golgi labeling combined with sevenfold expansion microscopy demonstrates BSC's ability to visualize dense neuronal morphology in clinically relevant human tissue.

Why it matters. The dataset reveals separated neuronal somata and their fine processes within densely labeled cortical tissue (Figure 1), providing morphology-rich reference data for AI segmentation, connectome reconstruction, cross-sample comparison, and human brain atlas development.

Making the brain computable across scales
From Sample to Computable Brain

A bidirectional physical–AI workflow.

This architecture is deliberately bidirectional: high-resolution imaging and molecular measurements provide ground truth for AI, while AI identifies objects, regions, and features, and generates hypotheses that can be returned to imaging, pathology, and clinical experts for validation. The result is an iterative physical–AI workflow rather than a one-way image-analysis pipeline.

01
Tissue & Biobanks
02
MExT & 3D Imaging
03
Spatial Omics
04
AI / HPC Reconstruction
05
Atlas, Disease & Care
More Specimens → More Data → Better AI → Greater Scientific Value
Core Technological Pillars

Four pillars powering next-generation neuroscience.

01

Advanced MExT Platform

Overcomes axial resolution limits in large-volume tissue microscopy, delivering high-throughput and near-isotropic 3D reconstructions across multi-centimeter biological samples.

02

Cross-Scale Resolution Propagation

Leverages high-resolution ultrastructural reference data to guide AI-assisted feature inference across entire tissue blocks, linking micro-architecture to organ-scale contexts.

03

Quantitative Brain Models

Converts static histological images into versioned, standardized digital assets — combining aligned volumetric data, automated segmentation, spatial omics, and connectomic graphs.

04

Comparative Neural Computation

Extracts principles of memory, sensory processing, and decision-making from experimentally tractable Drosophila circuits to inform human cognitive architecture and neuromorphic AI.

Selected Publications

From molecules and synapses to circuits, cognition, and care.

01
Lead impairs long-term memory by disrupting mitochondrial redox homeostasis in Drosophila.
Proceedings of the National Academy of Sciences (in revision)
2026
02
Individualized decision making driven by long-term memory retrieval in Drosophila.
Cell Reports
2026
03
Self-supervised isotropic resolution enhancement of expansion microscopy via quantized compression.
Journal of Imaging Informatics in Medicine
2026
04
Dopaminergic modulation of mushroom body output neurons mediates nociception-induced escape in Drosophila.
eLife
2026
05
Hybrid neural networks in the mushroom body drive olfactory preference in Drosophila.
Science Advances
2025
06
Redox regulation of memory formation by Rrp1 in Drosophila.
Proceedings of the National Academy of Sciences
2025
07
Rapid single-wavelength light-sheet localization microscopy for clarified tissue.
Nature Communications
2019
08
Toward whole-body connectomics.
Journal of Neuroscience
2016
09
Parallel neural pathways mediate CO₂ avoidance responses in Drosophila.
Science
2013
10
Visualizing long-term memory formation in two neurons of the Drosophila brain.
Science
2012
11
A map of olfactory representation in the Drosophila mushroom body.
Cell
2007
Social Impact

From brain atlases to biomedical intelligence.

BSC's primary outputs are not images alone, but quantitative and reusable models: connectome maps, molecularly annotated three-dimensional atlases, validated AI models, disease-specific spatial signatures, and computable representations of individual brains.

These resources will reveal vulnerable cell populations, clarify circuit-level mechanisms, identify therapeutic targets, strengthen interpretation of clinical specimens, and provide a structural foundation for precision diagnosis and treatment.

The scale of brain data also creates a frontier for computer science. Reliable analysis of terabyte- and petabyte-scale datasets requires advances in automated acquisition, data movement, distributed inference, uncertainty-aware reconstruction, and human–AI collaboration. In return, decoding the brain's efficient and robust computational architecture may inspire more energy-efficient and trustworthy artificial intelligence.

🧬
Biomarker Discovery
💊
Drug Target Identification
🔬
Disease Classification
🧠
Digital Brain Models
📊
Prognosis Prediction
🎯
Precision Treatment
“

BSC aims to become a globally connected platform for human brain mapping and AI-enabled precision medicine. By uniting neuroscience, imaging science, engineering, computation, and clinical practice, the Center will make brain structure navigable across scales — revealing how the brain computes, how disease disrupts its organization, and how this knowledge can improve human health.

Our Vision
Contact

Get in touch with the Center.

Shuinan Main Campus

Telephone
04-2205-3366
Safety Line
04-2202-2205
Email
bsc@mail.cmu.edu.tw
Address
No. 100, Section 1, Jingmao Rd.,
Beitun Dist., Taichung 406040,
Taiwan
經貿路一段 · Jingmao Rd.
Our Location
Brain Spatiomics Center · CMU Shuinan Campus
No. 100, Section 1, Jingmao Rd., Beitun Dist., Taichung 406040
Open in Google Maps