Core Contributors
Faculty
I want to quantify glymphatic flows and understand them in terms of physical mechanisms like pressure, stress, forces, momentum, and energy – at brain-wide and local scales. I want to use that understanding to suggest behaviors and interventions that can promote healthy glymphatic function and improve clinical outcomes. I want to link physical mechanisms to the neurological processes that control, and are controlled by, them. Tight coupling between models and experiments drives us forward.
Techniques:
- Automated particle tracking
- Automated front tracking
- Image processing
- Physics-informed neural networks
- Reduced-order modeling
- Glymphatic flow simulation & theory
- Fluid dynamics experiments
Our aim is to quantify glymphatic flow and transport using data analysis, mathematical models, and simulations to (1) augment experimental data and (2) provide insights into glymphatic transport and how it is altered under various conditions.
Techniques: Image processing, including particle tracking, image segmentation, vessel diameter, pulsatility, and front tracking, data analysis, reduced-order mathematical modeling, numerical simulations
Fluid dynamics of cerebrospinal fluid flow in the brain; advective and diffusive transport of solutes in the brain
Techniques: Theoretical modeling of the glymphatic system, analytical and computational solutions of the governing equations of fluid dynamics, advection, and diffusion, interpretation of experimental data on glymphatic transport.
Postdoc
CSF, ISF flow meeting, arterial-pumping flow simulations
We build fluid dynamic models to understand the driving (pumping) mechanism of CSF flow in the PVS. Discover the potential role of astrocyte endfoot, functional hyperemia, and cardiac pulsation in driving the flow. In particular, we want to know how the gaps between anstrocyte endfoot may act as valves that rectify the CSF-ISF exchange.
Techniques: Fluid-Solid Interaction Simulation, Image processing, particle tracking.
Staff
Our aim is to understand how astrocyte endfeet may function as valves to rectify CSF flow, leading to observed net flow in experiments. Current work utilizes three-dimensional fluid-structure interaction modeling to investigate possible endfeet valve mechanisms, and we quantitatively examined how CSF flow is rectified by asymmetry in endfeet geometry. The parameters we are examining include endfoot thickness and length, gap size, material properties of astrocyte cells, oscillatory pressure in PVS and pressure frequency.
Techniques: Computational fluid dynamics, Computational solid mechanics, Fluid-structure interaction simulation, Mathematical modeling
PhD Students
Quantification of Glymphatic System using Scientific Machine Learning and Numerical Simulations
We aim to quantify and determine different physical parameters in the glymphatic system of the brain. To achieve this, we are using three types of experimental data: 1) In-vivo two-photon microscopy data, 2) Transcranial imaging data, and 3) DCE-MRI data. Each dataset has its own advantages, but to quantify the CSF flow, we need to calculate certain physical quantities based on these raw data. We use Scientific Machine Learning methods, specifically Physics-Informed Neural Networks (PINNs), to reconstruct the flow field, enabling us to compute various fluid dynamics quantities. This approach, also known as Artificial Intelligence Velocimetry (AIV), allows us to combine both experimental data and governing physical equations to obtain more accurate and physically interpretable results. Every AI model requires validation. To validate our models, we simulate the fluid dynamic system using a fluid dynamics solver and treat the known results as synthetic data.
Techniques: Machine learning, Computational Fluid Dynamics, Image Processing, Data analysis, Mathematical Modeling.
Experimental Validation of Pores as Valves
Our aim is to identify and model potential drivers of flow in the perivascular space (PVS). Our current focus is on the potential of gaps between astrocyte endfeet to act as valves, driving the flow along the PVS in one direction. To examine the validity of this mechanism, we're creating membranes with asymmetric pores out of silicone and examining how flow through the pore changes with pressure.
Techniques: Lab scale device fabrication (silicone casting, machining), simulation (FEBio, SolidWorks)
Modeling of brain-wide solute transport and arterial-pulsation-driven flows
Our aim is to understand brain-wide fluid and solute transport through networks of pereivascular spaces and the adjacent parenchyma. Analytical modeling requires simplifications, but the networks can be solved rapidly. Efficient concentration calculations enable repeated simulations across estimated ranges of geometrical parameters for these spaces. The models are verified by comparison to in vivo measurements of fluid and solute transport in pial perivascular spaces. We use particle tracking velocimetry and image analysis techniques on two-photon images to study flows and perivasular geometries in vivo.
Modeling Techniques: Hydraulic networks, analytical steady-state advection-diffusion equations. Data Analysis Techniques: Particle tracking velocimetry, Z-stack segmentation, vessel diameter measurements, vessel pulsatility analysis
Performing vessel diameter measurements on time series images using various algorithms, including those focused on edge detection (FIE), image intensity (VD), and the Radon transform (RT-FWHM). Utilizing the front tracking algorithm to analyze the velocity field of cerebrospinal fluid (CSF) at different brain locations.
Undergraduate Researchers
Our aim is to quantify flow in various biofluids using various forms of imaging. One project aims to develop a method of obtaining three-dimensional velocity fields of lymphatic vessels using physics-informed neural networks from two-photon microscopy imaging. I have also worked on creating synthetic cardiac vessel data to analyze vessel diameter algorithms. Lastly, the current work focuses on quantifying the velocities associated with tracer-injected spinal cords in mice MRI imaging.
Techniques: Particle tracking velocimetry, front tracking, image segmentation, image registration