Ph.D. Oral Defense: "Direct Reconstruction of binding potential for positron emission tomography" [Mr. Paul Gravel]
Ph.D. ORAL DEFENSE - Department of Biomedical Engineering
Mr. Paul Gravel:
Direct Reconstruction of binding potential for positron emission tomography
Friday, November 21, 2014 at 9:15 a.m.
Duff Medical Bldg.
Conference Room 333
3775 University Street
(Enter via University Street. Continue straight through the Lobby and down the corridor, turn right just before the double doors. Take the elevator or stairs up to the third floor. Room 333 is at the end of the hallway)
ABSTRACT
Currently, positron emission tomography (PET) produces reconstructed images that require considerable post-processing efforts, in order to deliver meaningful results to researchers. This work proposes to alleviate these post-processing efforts by incorporating each step within the reconstruction framework of the raw PET data to provide the researchers directly with better images of interest, in a stereotaxic space for easy interpretation.
In neuroscience and psychiatry, the goal of many studies using PET is to assess whether or not specific biological parameters significantly differ between groups or conditions. However, to get the biological parameters of interest, tracer kinetic modeling techniques are required . These techniques are conventionally appl ied after reconstruction of the PET data. In addition, to test for significance at the voxellevel statistical parametric mapping is often used which requires all the images to be registered in a common spatial atlas, or stereotaxic space. Typically, this step is performed after PET image reconstruction as well.
It is becoming increasingly clear that it is often better, in terms of achieving better quantification , to estimate the parameters of interest directly from the raw Poisson distributed PET data, rather than to first reconstruct and then separately post process the data.
Therefore, the first aim of this project was to incorporate the registration to anatomical MR (or stereotaxic) space into the image reconstruction algorithm and assess its performance by comparing the results with those of conventional post-reconstruction registration. The second aim was to combine the simplified reference tissue model with the basis function method (SRTM-BFM) tracer kinetic model with the recently developed direct 40 PET one-step late maximum likelihood expectation maximization (OSL-MLEM) reconstruction algorithm and evaluate its performance with the conventional post-reconstruction method. Finally, the methods developed and assessed for aims 1 and 2 were combined and applied on a [11C]raclopride-like simulated study consisting of two conditions, to test whether or not statistical significance is increased in comparison to the conventional post-reconstruction method.
The proposed research has the potential to improve current PET method(s) and constitutes an important step towards the goal of providing researchers with , not only better results, but directly interpretable results.