Thesis Defense - Arthur Mikhno
Non-invasive and cost-effective quantification of Positron Emission Tomography data
Date and Time: Tuesday, October 28th at 4pm
Location: CEPSR 414
Sponsor: Prof. Andrew Laine, PhD
Molecular imaging using Positron Emission Tomography (PET) is beginning to revolutionize drug development, drug delivery targeting, early diagnostics, and screening for clinical trials. PET images cannot be fully quantified without arterial blood sampling during the scan. Arterial blood sampling is invasive, risky, costly, time consuming and uncomfortable for the patient. While many approaches have been tried, the ultimate goal of zero blood samples has remained illusive for over a decade. In the dissertation we break this proverbial blood barrier and present for the first time a totally non-invasive PET quantification framework. This is accomplished with a combination of novel image processing, modeling, and tomographic reconstruction tools.
First, we developed dedicated pharmacokinetic modeling, machine learning and optimization framework based on the fusion of Electronic Health Records (EHR) data with information from dynamic PET brain images. This allows us to non-invasively infer a patient’s metabolism and clearance rates of the PET tracer from the body.
Next, a new tomographic reconstruction technique is proposed that combines locally weighted denoising with a new resolution modeling approach. This technique allows visualization and quantification of structures smaller than previously possible.
This work brings us closer to the ultimate goal of reducing the cost and complexity of PET, which will facilitate translation of many research tracers for clinical use.