1. NeuroInformatics Center
Neuroinformatics and High-Performance Grid Computing for Rural Telemedicine
Allen D. Malony
Computational Science Institute
Department of Computer and Information Science
Don Tucker
Electrical Geodesics, Inc.
Department of Psychology
NeuroInformatics Center
University of Oregon

2. Outline Neuroinformatics research
Dynamic brain analysis problem
NeuroInformatics Center (NIC) at UO
Neuroinformatics technology and applications
Dense-array EEG and Electrical Geodesics, Inc. (EGI)
Epilepsy and pre-surgical planning
NIC research and development
High-performance and grid computing
ICONIC Grid HPC system at UO
HPC/Grid computing for Oregon’s health care industry
E-Health grid for rural telemedicine

3. Integrated Dynamic Brain Analysis
Individual Brain Analysis
Structural /
Functional
MRI/PET
Dense
Array EEG / MEG
Constraint
Analysis
Head Analysis
Source Analysis
Signal Analysis
Response Analysis
Experiment
subject
temporal
dynamics
neural
constraints
Cortical
Activity Model
Component
Response Model
spatial pattern
recognition
temporal pattern
Cortical Activity
Knowledge Base
Component Response
good spatial
poor temporal
poor spatial
good temporal
neuroimaging
integration
In cognitive neuroscience studies, it is has been shown that EEG and fMRI studies provide complimentary information that informs researchers about brain functions. We believe that the EEG can be used to compliment other brain imaging methods.

4. Experimental Methodology and Tool Integration
16x256 bits per millisec
(30MB/m)
CT / MRI
EEG
segmented tissues
NetStation
BrainVoyager
processed EEG
mesh generation
source localization constrained to cortical surface
Constraining the source solutions to the cortical surface is a major advantage for analyzing EEG and ERP effects. A working assumption is that sources are likely to be oriented normal to the cortical surface.
Interpolator 3D
EMSE
BESA

5. NeuroInformatics Center (NIC) at UO
Application of computational science methods to human neuroscience problems
Tools to help understand dynamic brain function
Tools to help diagnosis brain-related disorders
HPC simulation, large-scale data analysis, visualization
Integration of neuroimaging methods and technology
Need for coupled modeling (EEG/ERP, MR analysis)
Apply advanced statistical analysis (PCA, ICA)
Develop computational brain models (FDM, FEM)
Build source localization models (dipole, linear inverse)
Optimize temporal and spatial resolution
Internet-based capabilities for brain analysis services, data archiving, and data mining

6. Electrical Geodesics Inc. (EGI)
EGI Geodesics Sensor Net
Dense-array sensor technology
64/128/256 channels
256-channel geodesics sensor net
AgCl plastic electrodes
Carbon fiber leads
Net Station
Advanced EEG/ERP data analysis
Stereotactic EEG sensor registration
Research and medical services
Epilepsy diagnosis, pre-surgical planning

7. Epilepsy
Epilepsy affects ~5.3 million people in the U.S., Europe, & Japan
EEG in epilepsy diagnosis
Childhood and juvenile absence
Idiopathic (genetic)
“Generalized” or multifocal?
EEG in presurgical planning
Fast, safe, inexpensive
128/256 channels permit
localization of seizure onset
Requires massive storage and data analysis capabilities
Unifying Features of Idiopathic Generalized Epilepsies
•Normal neurologically before onset of seizures
•EEG changes appear generally well organized with otherwise normal background
•Long term outlook for cognitive function is good
Many of the epilepsies previously classified as “generalized” are not truly generalized.
•Idiopathic Generalized Seizures occur in neurologically normal children
•Most seizures or epilepsy syndromes previously classified as “generalized” or “undetermined” should be thought of as Epileptic Encephalopathies
•Epileptic Encephalopathies may have an identifiable mechanism or respond to specific therapies

8. EEG Methodology Electroencephalogram (EEG) EEG time series analysis
Event-related potentials (ERP)
averaging to increase SNR
linking brain activity to sensory–motor, cognitive functions (e.g., visual processing, response programming)
Signal cleaning (removal of noncephalic signal, “noise”)
Signal decomposition (PCA, ICA, etc.)
Neural source localization

9. EEG Time Series - Progression of Absence Seizure
First full spike–wave
Figure 1. Topographic waveform plot for the first full spike-wave in one patient’s seizure. Following the clinical EEG convention, negative is up. The 256 channels are arrayed in two dimensions as they would be seen looking down on the top of the head, with the nose at the top of the page, and with the lower channels unwrapped to the sides of the page. This 350 ms epoch captures the first wave and spike of this seizure, plus the initial onset of the second wave. Note that the prominent spike-wave pattern over medial, superior frontal sites inverts over lateral, inferior frontal sites, indicating neural sources in medial frontal cortex.

10. Topographic Mapping of Spike-Wave Dynamics
Spatial and temporal dynamics need to be observed
1 millisecond temporal resolution is required
Spatial density to capture shifts in topography
Linked networks contribute to measured potentials
Problem of superposition (source number and location)
Figure 2. Selected topographic maps for the spike-wave pattern shown in Figure 3. Palette is scaled for this wave and spike interval, from –130 uV (dark blue) to 75 uV (dark red). The interval characterized by each map is 1 ms, with the selections made to illustrate the major topographic transitions of this subject’s spike-wave pattern.

11. Dipole Sources in the Cortex
Scalp EEG is generated in the cortex
Interested in dipole location, orientation, and magnitude
Cortical sheet gives possible dipole locations
Orientation is normal to cortical surface
Need to capture convoluted geometry in 3D mesh
From segmented MRI/CT
Linear superposition

12. Building Computational Brain Models
MRI segmentation of brain tissues
Conductivity model
Measure head tissue conductivity
Electrical impedance tomography
small currents are injected between electrode pair
resulting potential measured at remaining electrodes
Finite element forward solution
Source inverse modeling
Explicit and implicit methods
Bayesian methodology

13. ICONIC Grid at the University of Oregon
graphics workstations
interactive, immersive viz
other campus clusters
Internet 2
Gbit Campus Backbone
NIC
4x8
CIS 16
CIS 16
CNI
2x8
NIC
2x16
SMP
Server
IBM p655
Shared
Memory
IBM p690
Graphics
SMP
SGI Prism
Distributed
Memory
IBM JS20
Distributed
Memory
Dell Pentium Xeon
Tape Backup
SAN Storage System IBM SAN FS
5 Terabytes

14. ICONIC Grid Hardware IBM p690 IBM p655 IBM FAStT Dell cluster IBM JS20
 16 processors
IBM p655
 4 nodes
 8 processors per node
Fibre Channel
Fibre Channel
IBM FAStT
 5 TB
 SAN FS
Dell cluster
 16 nodes
 2 processors per node
IBM JS20
 16 blades
 2 processors per node

15. Computational Integrated Neuroimaging System…
raw
storage
resources
virtual
services
compute resources

16. Leveraging Internet, HPC, and Grid Computing
Telemedicine imaging and neurology
Distributed EEG and MRI measurement and analysis
Neurological medical services
Shared brain data repositories
Remote and rural imaging capabilities
Enhanced HPC / grid infrastructure for neuroinformatics
Build on emerging web services and grid technology
Link HPC servers with high-bandwidth networks
Create institutional and industry partnerships
Cerebral Data Systems (UO partnership with EGI)
Continue strong relationship with IBM and Life Sciences

17. Oregon E-Health Grid for Rural Telemedicine
Region 2
Internet 2 / National LambdaRail
Region 1
Regional networks
Region 5
Region 4
HPC servers
Rural sites
Region 3
Regional clients