1. conflicts of interest to report.
Conflict of Interest Disclosure Amanda F. Khan, MSc. Medical Biophysics
Has no real or apparent
conflicts of interest to report.

2. A Novel MRI-Compatible Brain Ventricle Phantom for Software Validation
Amanda F. Khan
Department of Medical Biophysics
Imaging Research Laboratories
Robarts Research Institute
The University of Western Ontario
Supervisors: Dr. Michael Borrie, Dr. Robert Bartha
Alzheimer’s Disease International – March 27h, 2011

3. The Ventricular System
Lateral ventricles
structures containing CSF in the midbrain
atrophy of surrounding tissues leads to increase in CSF volume
increase in CSF = increase in lateral ventricles (surrogate measure)
capture this increase on MRI, sometimes years before cognitive decline can be measured
Control AD
3D image adapted from: The Biodidac; MRI images from Wahlund LO, et al., J Neurol Neurosurg Psychiatry. 2000;69(5):630–635.

4. Brain Atrophy and Alzheimer’s
How do we objectively quantify brain atrophy?
How can we track ventricular expansion?

5. Volumetry and Alzheimer’s
Segmentation programs
objectively measure changes in ventricle volume = surrogate of brain atrophy
uses seed placement and atlas systems
potential to evaluate neuron saving potential of new AD drugs in clinical trials
have not been reliably validated using physical models
Control AD

6. Overview of the Phantom Model
Purpose: to create a standardized, MRI compatible life-size phantom model of the human brain ventricles to evaluate segmentation methods utilizing 3T T1-weighed MRI 1 2 3 4

7. Ventricle Construction
Pre-processed image
Post-processed image
ADNI 3T MRI image
ADNI Image Database
Software images from: The University of Pennsylvania

8. Ventricle Construction
Digital mesh
ITK-SNAP
Rapid-prototyped lateral ventricles
Fused Deposition Modelling

9. Tissue Model
Series Description:
MP-RAGE ADNI iPAT2 Axial
Slice Thickness: 1.2mm
Repetition Time: 2300ms
Echo Time: 3ms
Number of Averages: 1
Inversion Time: 900ms
Flip Angle: 9˚
Width: 256mm
Height: 240mm
ROI Brain
2% agar
ROI Ventricle
Signal intensity (SI) difference = SI brain tissue – SI ventricle

10. Tissue Model Series Description: MP-RAGE Slice Thickness: 1.2mm
Repetition Time: 2300ms
Echo Time: 3ms
Number of Averages: 1
Inversion Time: 900ms
Flip Angle: 9˚
Width: 256mm
Height: 240mm

11. Brain Ventricle Phantom
300mL “base” for proper midbrain positioning
Completed brain-ventricle phantom model
Slice Thickness: 1.2mm, Repetition Time: 2300ms, Echo Time: 3ms, Number of Averages: 1, Inversion Time: 900ms
Flip Angle: 9⁰, Acquisition Width: 256mm Height: 240mm

12. Results Signal Intensity vs. Gadolinium Concentration
Slice Thickness vs. Volume Accuracy

13. In-Vivo 3T Patient Scans
Results
Signal Intensity Properties
In-Vivo 3T Patient Scans
Phantom 3T Scans
261.5 SI units
262.9 SI units
Volumetry Accuracy
True Volume
Phantom 3T Scans
47.7 cm3
47.2 cm3
3T MRI scans of completed phantom

14. Discussion cost –effective (under $400 for materials)
preserving agents / refrigeration can be used to prolong shelf-life
creation is simple and can be replicated by any other group
methods can be modified to suit unique research areas
is an appropriate method to validate segmentation methods

15. Acknowledgements Supervisors: Sources of Funding and Collaboration:
Dr. Robert Bartha
Dr. Michael Borrie
Collaborators:
John Drozd
Robert Moreland
Robert Ta
Support:
Kim Krueger
Adam McLean
Joe Gati
Michael Marynowski
Henry Betta
Vaishali Karnik
Sources of Funding and
Collaboration: