On The Feasibility of Positron Emission Tomography Imaging Systems Utilizing Mixed Spatial Resolution Detectors

Positron emission tomography (PET) is a powerful molecular imaging modality used to address a wide assortment of pathologies, including cancer, neurodegenerative diseases, psychiatric conditions, and cardiac disease. It is capable of providing specific and quantitative information about biological processes, but this capability is dependent on its ability to resolve regions of interest in the body accurately. In recent years, while preclinical and research systems have pushed spatial resolution capability toward sub-millimeter levels, improvements in spatial resolution performance for clinical PET systems have plateaued at around 4 mm due to practical constraints on cost and engineering complexity. However, resolution performance better than 4 mm is desirable to multiple applications in brain, breast, prostate and pediatric PET imaging among others. Therefore, alternative strategies to boost spatial resolution performance in at least part of the imaging field of view (FOV) are of interest. %pose the problem The work presented examined the feasibility of replacing some fraction of a standard resolution scanner's detectors with higher spatial resolution detectors as a method to improve spatial resolution performance, generating data containing a mix of the resolution properties of the different detectors. Two scanner configurations were investigated: (1) a scanner configuration with a centrally located region of high resolution detectors and (2) a scanner configuration with an end-located region of high resolution detectors. %add another sentence here In order to account for the mixed intrinsic resolution properties of the collected data from these geometries, a three-dimensional iterative maximum likelihood expectation maximization reconstruction method was developed and validated. The algorithm employed a ray-tracing projection method using CPU multithreading for speed up in order to flexibly calculate system matrices for the different scanner geometries studied. An image-space resolution model created from sparse axial sampling of the point system responses along the axial FOV was used to account for the variable spatial resolution caused by the mixed resolution data. Reconstructions using this model demonstrated improved resolution performance compared to those without. Each of the mixed resolution scanner configurations examined were compared to scanners using a single type of spatial resolution detector, either high or standard. The standard resolution systems had the same overall geometry as the mixed resolution systems but had uniformly lower resolution detectors and was used to evaluate the improvements in spatial resolution performance of the mixed resolution systems. In order to understand the sensitivity benefits of additional but lower resolution lines of response of the mixed resolution systems, the compared high resolution systems had the same geometry as the high resolution detector module region of the corresponding mixed resolution system. Studies evaluating sensitivity, spatial resolution performance and contrast recovery showed advantages of the mixed resolution systems. Initial work on an evaluation system using a transformable gantry was performed. Data acquisition software was developed for a four-module system of high resolution detectors arranged in a box-like geometry. Further development and characterization of this system is ongoing with the goal of validating the mixed resolution concept using real data. In summary, this work developed methods to study the feasibility of a mixed resolution PET scanner geometry and evaluated two candidate system geometries, determining that mixed resolution systems have potential benefits in terms of both sensitivity and spatial resolution performance. Initial work was performed developing an evaluation system for real data validation