Projective Methods for PET/SPECT Image Analysis

Yale University, New Haven CT

Investigators

Abstract

Project Summary/Abstract This research proposes a new image analysis technology for PET/SPECT neuroimaging. Analysis of PET/SPECT neuroimages often requires an image reference region where the radiotracer has non-specific binding. Unfortunately, many old as well as emerging radiotracers do not have a useful reference region, and this can limit their use. In this research, we propose the development of a new computational technology for processing such images using ideas from projective geometry. This methodology will enable reference-region- free analysis of PET/SPECT images, thereby enabling a more widespread use of tracers that do not have an optimal reference region. There are three specific aims to this exploratory research: The first is to fully develop the mathematics and algorithms of the projective approach. The second is to evaluate the proposed methods using a large image dataset of a SPECT tracer (Ioflupane) with a known reference region. This tracer is used in imaging Parkinson's Disease. This dataset will enable the comparison of the new methods (with the reference region ignored) with classical methods which use the reference region. Finally, we will evaluate the performance of this method with a new PET tracer for synaptic vesicle protein SV2A. This tracer has a suboptimal reference region; no non-specific binding region is known. The results of the reference region free method will be compared with classic compartmental modeling.