Aaron Fenster PhD, FCCPM : Home Page

Coherent-Scatter Computed Tomography

INTRODUCTION:

Identification of vulnerable plaques that are likely to rupture leading to throbogenic events is now a major challenge in the management of risk for stroke. In addition, improved strategies to treat atherosclerosis non-surgically require sensitive non-invasive 3D imaging techniques allowing direct plaque visualization for serial monitoring of disease progression and regression. Thus, the underlying hypothesis of our research program is that 3D ultrasound (US) imaging alone or in combination with 3D MR images of the carotid arteries will provide sensitive and quantitative tools allowing quantification of plaque vulnerability factors and allowing detailed quantification of plaque changes. The goal of this grant is to develop improved non-invasive 3D ultrasound imaging tools, which will allow measurement and visualization of plaque volume, surface features, and composition, and provide detailed information on changes in carotid plaques. This information will be correlated to 3D MR images of carotid plaques as part of the carotid Theme of our CIHR Group in Vascular Imaging. Our specific goals are:

GOAL (1): To develop a 3D plaque surface morphology segmentation technique

Algorithm: We will extend our deformable balloon model approach to the segmentation of the plaque surface. Our approach will involve 3 steps: interactive placement of an initial model inside the lumen; inflation of the model towards the plaque surface; and, automatic localization of the plaque surface.

Validation: The accuracy and variability studies will be based on the boundary density distribution approach we have introduced. Results from repeated automated and manual segmentations of multiple phantoms with various sized of ulcerations and edarterectomy specimens will be statistically analyzed.

Cardiac pulsation: Cardiac gating would eliminate variation of the vessel wall due to cardiac pulsation, but would result in long scanning times. We will develop a retrospective gating approach based on analysis of temporally over-sampled images. Slower 3D scanning will result in over-sampled data, which will be analyzed and rearranged to obtain 3D images of vessel at the different phases of the cardiac cycle.

GOAL (2): To develop plaque volume segmentation and characterization techniques

Plaque volume algorithm: We will use a region growing process based on image features to separate the plaque from its surrounding. These will be based on the statistical properties of the local pixel intensity distribution. The results will be integrated into the 3D visualization software we have already developed.

Validation: We will perform multiple repeated segmentations of plaque volumes in phantoms with various sizes and echogenicity of simulated plaques and edarterectomy specimens. Statistical analysis will give us the accuracy and variability of plaque volume segmentation using our algorithm. These will be compared to the inter- and intra-operator variability and accuracy obtained with multiple operators.

3D plaque composition analysis: We will pursue two approaches. Analysis of 3D B-mode images using co-occurrence matrices will be easier to implement on any ultrasound machine but will not separate lipid and thrombi pools. RF signal analysis will separate these constituents but will not be easily implemented in systems in general use. Comparisons will be made with endarterectomy specimens and with MRI.

GOAL (3): To develop 3D image registration techiques to allow monitoring of plaque changes

Algorithm: A mutual information-based technique will be developed to register 3D B-mode/MRI images and US with itself of the carotid plaques. This will allow validation and comparisons with MRI and permit serial studies of plaque progression/regression.

Validation: We will test the accuracy and robustness of the approach by misaligning images by known translations and rotations and then determining the accuracy by which fiducial points are reregistered.

Investigators: Fenster, A., Downey, D., Spence, D.
Postdoctoral Fellows: Cardinal, N.
Graduate Students: Gill, J., Patenaude B., Landry A.
Technicians: Blake, C.
Support: CIHR and ORDCF

----

E-mail: afenster@imaging.robarts.ca

Phone: (519) 663-3833 Fax: (519) 663-3900

Copyright 2005 ©  All rights reserved.