Biomedical Informatics Research Network Michael Marron Ph D
Biomedical Informatics Research Network Michael Marron, Ph. D. , Director Division of Biomedical Technology National Center for Research Resources National Institutes of Health Department of Health and Human Services Bethesda, Maryland, USA 1
Biomedical Informatics Research Network Testbed for biomedical knowledge infrastructure January 2003 2
BIRN Biomedical Informatics Research Network A testbed for a biomedical knowledge infrastructure Federated database of neuroimaging data Fusion of diverse data sources (location; level of aggregation) Grid access to computational resources Datamining software Scalable and extensible Driven by research needs-pull, not technology-push 3
BIRN Sites 4
BIRN Information Integration What is the cerebellar distribution of rat proteins with more than 70% homology with human NCS-1? Integrated View Definition Wrapper Mediator Wrapper Web protein localization morphometry neurotransmission Ca. BP, Expasy 5
BIRN Grid Infrastructure • The BIRN Portal is built on many “layers” • Layers are modular, allowing for extension of any layer without great disruption to the entire system • Every Layer has its own complexity and administration that was previously passed on to the end-user • Portal centralizes all administrative details of each layer into a single username and pass phrase January 2003 6
BIRN Grid Infrastructure BIRN Portal BIRN Toolkit Collaboration Applications Viewing/Visualization Data Management Queries/Results Grid. Port Toolkit Custom APIs Visualization APIs Security API Grid Middleware Mediator API SRB Globus Mediator Database MCAT Computation Distributed Resources Data Storage 7
BIRN Knowledge Integration Supports queries across heterogeneous data sources Mediator Services API Query Deductive Engine Mediator Layer Query Formulation Source Model Lifting Model Reasoner Query Processing Source Registration Optimizer Wrapper Layer Query interface Result interface SRB Files January 2003 RDB HTML OODB XML 8
Mouse BIRN Studying animal models of disease across dimensional scales to test hypothesis with human neurological disorders § Experimental Allergic Encephalomyelitis (EAE) mouse models (both chemically induced and transgenic) exhibit episodic weakness and demyelination characteristic of multiple sclerosis § Dopamine Transporter (DAT) knockout mouse for studies of schizophrenia, attention-deficit hyperactivity disorder (ADHD), Tourette’s disorder and substance abuse § Using an alpha-synuclein mouse to model the symptoms/pathology of Parkinson’s Disease Duke, UCLA, UC San Diego, Cal Tech January 2003 9
Mouse BIRN Integrating brain data across scales and disciplines Reconstructed spiny dendrite Coronal sections of mouse brains UCSD-NCMIR Duke - CIVM UCSD-NCMIR & Duke - CIVN January 2003 10
Morphology BIRN Examining neuroanatomical correlates of neuropsychiatric illnesses, including Unipolar Depression, mild Alzheimer’s Disease and mild cognitive impairment Combining data from multiple acquisition sites Increase statistical power for study of rare populations Harvard (MGH and BWH), Duke, UCLA, UC– San Diego, Johns Hopkins January 2003 11
Morphology BIRN Harvard-MGH Harvard-BWH Surface based coordinate system Model based three dimensional medical image segmentation UCLA-LONI Dynamics of gray matter loss rates, mapped in a schizophrenia population Harvard-BWH 3 D-Slicer - An Integrated visualization system for surgical planning and guidance using image fusion and interventional imaging 12
Functional BIRN } Developing a common f. MRI protocol to study regional brain dysfunction related to the progression and treatment of schizophrenia } Investigating techniques to insure interoperability of existing tools for multi-modal analysis } Correlating functional data with anatomical data acquired from the Morphology test-bed to study if there are neuroanatomical correlates with cognitive dysfunction across disorders } UCLA, UC San Diego, UC Irvine, Harvard (MGH and BWH), Stanford, Minnesota, Iowa, New Mexico, Duke/U. North Carolina January 2003 13
Functional BIRN Harvard-MGH Patients with schizophrenia show abnormal modulation of temporal and frontal regions during semantic processing neocortical network mediates semantic processing of sentences in healthy individuals f. MRI response at 3 T and 1. 5 T with identical software and hardware platforms January 2003 Stanford – Lucas Center 14
BIRN Coordinating Center } Deploy a network infrastructure capable of quickly moving large amounts of data between BIRN sites across the country } Create federated databases pertaining to the BIRN scientific test-beds } Develop software to find, compare, and analyze complex neurological imaging data } Ensure regulatory compliance (HIPAA) without inhibiting collaboration } UC San Diego January 2003 15
BIRN Network Operations Center Standardized Site Rack 24 x 7 Operations Monitoring BIRN NOC Monitoring Internet 2 UCSD/SDSC Network NCRR GB Switches • Network Switch Cisco 4006 • BIRN Statistics DL 380 Network Stats • BIRN Network Probe Gig. E Net Probe • Network Attached Storage • Grid POP/SRB, Globus • General purpose ( eg. , encryption) • UPS for Rack N 2400 NAS 1 - 10 TB Gigabit Network Probes Wired and Wireless DL 380 Grid POP DL 380 APC UPS BIRN Site Rack January 2003 BIRN Network Management NOC Servers and Testers/Analyzers Ultimately, access from anywhere to BIRN data everywhere 16
BIRN CC Services } } } } } January 2003 Network Monitoring Statistics & Measurement 7 x 24 Help Desk Problem Tracking High Level Project Management Portal Services & Tools Integration Training Data Integration Visualization Tools Documentation – web site, best practices, lessons learned, checklists 17
Challenges Large data sets § ISCAR - information storage, curation, archiving, and retrieval § Bandwidth issues § Quality control of data only analyzed by machine § Visual interfaces for complex data sets Fostering collaboration § Reward structures in government and academic institutions § Intellectual property issues § Data integrity § Resistance to collaborate with competitors Data sharing issues (ownership, privacy, authorship) January 2003 18
More Challenges Federated DB /Knowledge Engineering § Dictionaries & Ontologies § Integrating different data types § Accessing legacy systems § Distributed query computations High-level management of projects § Overcome technical & institutional barriers § Degree of centralization needed § Role of standards § Project subdivision and optimization January 2003 19
BIRN Future } } } January 2003 Addition of new sites in a rapid, well-defined, affordable fashion Grid-based National Clinical Research Network Cross-species integration of data Characterization of treatment at multiple levels Expand model to other areas of biomedical research 20
- Slides: 20