Morphometry BIRN: Using structural MRI to advance our ' understanding of disease

1
Morphometry BIRN Using structural MRI to
advance our .understanding of disease

• Christine Fennema-Notestine, Ph.D.
• University of California, San Diego

2
Structural Magnetic Resonance Imaging
3
Morphometry BIRN Goal

• To enable multi-site clinical neuroimaging
  studies investigating
• neurodegenerative disease onset and progression
• neuroanatomical correlates of clinical
  performance and disease state
• effectiveness of therapeutic intervention
• Specifically within this testbed to examine
• Alzheimers disease (AD)
• Mild cognitive impairment
• Effects of normal aging

4
Morphometry BIRN Sites and Methods

• Nine currently active mBIRN sites
• MGH, BWH, Duke, UCLA,
• UCSD, UCI, JHU, Wash U, MIT
• Areas of Research
• MRI acquisition and calibration
• Enable data sharing
• De-identification of data
• Distribute and guide development of MRI analysis
  tools
• Integrate advanced image analysis and
  visualization tools

5
Advances in Structural MRI

• Image acquisition
• Traditional
  pulse
  
  sequences

• 
  
• Novel pulse
  sequence improves
  
  subcortical region

Fischl et al., 2006
6
Advances in Structural MRI

• Image calibration
• Distortion correction
  (1.5T)

Jovicich et al., 2006
7
Morphometry BIRN Sites and Methods

• Nine currently active mBIRN sites
• MGH, BWH, Duke, UCLA,
• UCSD, UCI, JHU, Wash U, MIT
• Areas of Research
• MRI acquisition and calibration
• Enable data sharing
• De-identification of data
• Distribute and guide development of MRI analysis
  tools
• Integrate advanced image analysis and
  visualization tools

8
Data Sharing and De-identification

• To comply with HIPAA regulations, remove
  identifying information from
• Image headers (e.g., name, age)
• Image volumes (i.e., remove facial features)
• BIRN De-identification Upload Pipeline (BIRN-DUP)

MGH/BWH/UCSD
9
BIRN-DUP De-identification Review
10
Morphometry BIRN Sites and Methods

• Nine currently active mBIRN sites
• MGH, BWH, Duke, UCLA,
• UCSD, UCI, JHU, Wash U, MIT
• Areas of Research
• MRI acquisition and calibration
• Enable data sharing
• De-identification of data
• Distribute and guide development of MRI analysis
  tools
• Integrate advanced image analysis and
  visualization tools

11
Novel Techniques
Subcortical Segmentation


12
Enabling Multi-Site Studies

• Scientific goal test new hypotheses on larger
  cohorts, to increase power to detect subtle
  changes
• Methodological goal to control variability
  across sites as much as possible
• Hardware
• Vendors differ in implementation (GE, Siemens,
  etc.)
• Type of head coil
• Software operating systems
• Pulse sequence specification
• Field strength of magnet (e.g., 1.5T, 3T)

13
Uncorrected
Corrected
1.5T
Jovicich et al., 2006
14
Enabling Multi-Site Studies

• Scientific goal test new hypotheses on larger
  cohorts, to increase power to detect subtle
  changes
• Methodological goal to control variability
  across sites as much as possible
• Hardware
• Vendors differ in implementation (GE, Siemens,
  etc.)
• Type of head coil
• Software operating systems
• Pulse sequence specification
• Field strength of magnet (e.g., 1.5T, 3T)
• Image analysis methods

15
Image Analysis Tools

• Subcortical segmentation

Fischl et al., 2002
16
Image Analysis Tools

• Cortical thickness

Gray-white boundary
Pial surface
Dale et al., 1999 Fischl et al. 1999 2000
17
Image Analysis Tools

• Cortical parcellation

• .

Fischl et al., 2004 Desikan et al., In Press
18
Image Analysis Tools

• White matter atlas
  
• Diffusion tensor imaging

MGH/BWH/JHU
19
Visualization

• For analysis,
  review of data/QA,
  
  neuroanatomical
  modeling

BWH/Pieper et al. http//www.slicer.org/

20
Proof of Concept Multi-site Study of Legacy Data

• Many research groups have valuable existing
  (legacy) data
• Can legacy data that varies with MR scan
  acquisition parameters be meaningfully reanalyzed
  as a larger combined dataset?
• Shared data from 1.5T sites (UCSD, MGH/BWH,
  WashU) for individuals over 60, some with AD
• Replicate expected, previous findings
• Examine best statistical approach to model
  multi-site data

21
Image Analysis

• Common image processing path
• Focused on age-related changes in hippocampal
  volume

Fischl et al., 2002
22
Replication of Previous Work

• Hippocampal volume loss in normal aging

Fennema-Notestine et al., 2005
23
Image Analysis

• Focused on hippocampal and temporal horn volumes

Fischl et al., 2002
24
Replication and Extension of Previous Work

• Diagnostic classification of healthy elderly and
  AD

Age-normalized Temporal Horn Volume
Age-normalized Hippocampal Volume
25
Prospective Multi-Site Studies

• New data collection
• standardized pulse sequences
• distortion correction, and
• common image analysis methods
• Large scale studies combining neuroimaging,
  clinical, and genetic data
• Vietnam Era Twin Study of Aging (VETSA)
• Morphometric and diffusion imaging to better
  understand genetic correlates of brain changes
• Alzheimers Disease Neuroimaging Initiative
  (ADNI)
• Over 40 sites to guide the development of MRI
  biomarkers for AD and treatment effects

26
Where to Find Us (nbirn.net)