Research

What we study

Our research turns multimodal brain images into biomarkers that capture resilience, aging, and disease — and into methods that make those biomarkers reliable and interpretable.

Default Mode Network (DMN24) seed-based group spatial map — sagittal and axial views

Resting-state functional networks

Characterizing brain functional organization from resting-state fMRI, including fine-scale connectivity and network atlases in aging and neurodegeneration, and their links to cognitive performance.

Amyloid PET: sagittal and axial standardized uptake value ratio maps

Imaging biomarkers of neurodegeneration

Developing and validating biomarkers for Alzheimer's and Parkinson's disease, including connectivity, amyloid-related, and shape-based measures.

Axial FLAIR brain with white-matter hyperintensities highlighted

Brain aging & resilience

Modeling how vascular risk, exercise, and lifestyle relate to cognitive resilience and brain maturity across the lifespan.

Support vector machine schematic: two classes separated by a maximum-margin hyperplane

Methods: AI & causality

Machine learning for individual prediction, novel information-theoretic measures (e.g. omega entropy), and causality analysis such as Dimensional Complexity and causalized convergent cross-mapping.

Spartan football player overlaid on a brain network — sports neurology and concussion research

Sports neurology & concussion

Applying our imaging expertise to pressing health issues in the Spartan community — mapping recovery trajectories of brain networks after sports-related concussion to protect the long-term health of student-athletes.

Multi-modal imaging templates: T1 MPRAGE, T2 FLAIR, ASL/CBF, DTI FA/MD/FW

Multi-modal pipelines

High-throughput, GPU-accelerated pipelines spanning structural, functional, diffusion, and perfusion MRI plus PET — harmonized for large, multi-site studies.

Computing

From workstation to supercomputer

Our analysis scales from the lab's own workstation to MSU's high-performance computing center without changing tools. Our local systems are a Dell Precision 7875 (AMD Threadripper PRO 7975WX, 512 GB ECC RAM, NVIDIA RTX A6000, 4 TB NVMe) and an M1 Ultra Mac Studio with 128 GB RAM, backed by an 84 TB Synology array on 10 Gb/s networking.

We run large-scale jobs on MSU's Institute for Cyber-Enabled Research (ICER) HPCC, using containerized pipelines built on FSL, SPM, AFNI, MRtrix, FreeSurfer, and ANTs, plus our own GPU-accelerated Python/MATLAB code, giving us a specialized environment highly optimized for neuroimaging and image computing on the HPCC.

512GBworkstation ECC RAM

A6000local NVIDIA GPU

84TBfast local storage

ICERMSU HPCC for scale-out