|
Development of High Performance Solvers for Anisotropic Systems
The anisotropic diffusion equation is important to a
variety of modeling applications, including
diffusion through biological tissues,
heat and mass transfer in microelectronics.
We have developed and implemented a
time-dependent anisotropic
diffusion solver framework, which is tailored to the cerebral
circumstance and has been designed for both sequential and parallel
computing environments using MPI (Message Passing Interface).
White Matter Fiber Tracking with Diffusion Tensor MRI
The information interpreted from diffusion tensor magnetic resonance imaging
(DT-MRI) technology lays the ground
for tracking fiber pathways and building connectivity maps
in the human brain.
Fiber tractography based on DT-MRI
has raised promises for a better understanding of
the white matter anatomy, and
anatomical connectivity and functional coupling
among different regions of the brain.
We proposed novel fiber tracking techniques that rely on
conducting consecutive simulations of anisotropic diffusion
over the brain volume using
diffusion tensor datasets acquired from living human brains.
Tractography Analysis on
White Matter Fiber Intregrity in Mild Cognitive
Impairment
Diffusion tensor MRI studies have unveiled that
Alzheimer's disease (AD)
can significantly deteriorate the integrity of white matter.
Since mild cognitive impairment (MCI) is considered a transitional
stage between normal aging and AD,
it would be interesting to see whether the integrity of
white matter tracts is compromised in MCI individuals,
such that the level of breach in fiber integrity
might serve as a biological indicator to monitor MCI
for probable progression onto AD.
We have used a tractography analysis method with
the diffusion simulation based tracking algorithm
to explore the integrity of white matter fibers in MCI.
Computational Schemes for Solving Convection Diffusion Equation
and Bioheat Equation
Many transport phenomena and processes, including fluid flow, heat and
mass transfer, can be modeled by a general convection diffusion equation.
We have proposed and developed a fourth-order compact finite difference scheme on
face centered cubic grids and a corresponding multigrid method
for solving the two dimensional convection diffusion equation.
The Pennes' equation is the most widely used computational model
proposed to study the bioheat transfer in living human tissues,
which is very useful for accurately evaluating skin burn and
for designing and establishing thermal treatments and protections.
We developed a new two level second order finite difference scheme
for solving the one dimensional Pennes' bioheat equation.
A First Prototype of PyACTS
The ACTS Collection is a set of software tools that help developers
or programmers write high performance parallel codes for
their scientific applications.
In order to provide a uniform easy-to-use external interface to
existing ACTS tools and support ACTS users
to rapidly prototype their codes with the tools,
during a summer visit to the Lawrence Berkeley National Laboratory,
I designed and implemented a first prototype of PyACTS,
a Python-based interface to the tools in the ACTS Collection.
Selected Publications
Software
More ...
| 