TY - JOUR T1 - Characterization of Partial Volume Errors in Coronary Plaque PET Imaging JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 450 LP - 450 VL - 61 IS - supplement 1 AU - Adam Alessio AU - Joe Adams AU - Martin Lyngby Lassen AU - Piotr Slomka Y1 - 2020/05/01 UR - http://jnm.snmjournals.org/content/61/supplement_1/450.abstract N2 - 450Objectives: Partial volume errors (PVE) in PET imaging occur because of the limited spatial resolution of the system and the regularization of noise in images. Typically, these errors are characterized with phantoms such as the NEMA IQ phantom for features ranging from 10 to 40 mm in diameter. PET is increasingly used to image smaller features, such as coronary plaques where 18F-NaF is a strong marker for microcalcification. The coronary plaques are often in the size range of 1-4 mm across and 2-10 mm long, smaller than the current NEMA IQ phantom range. This work developed a physical phantom to characterize the PET imaging performance of small plaque features in order to determine limits of detectability and PVE correction methods. Methods: A custom phantom was designed consisting of a background chamber and a cylindrical support structure that supports 15 cylindrical features. The features have 5 different diameters [13, 8, 5, 2.7, and 1.7 mm] and can be filled with different activity concentrations. As an initial test, the phantom was filled with a 20 to 1; 10 to 1; and 5 to 1 activity concentrations in the feature compared to background and scanned on a GE DSTE PET/CT system (Waukesha, WI). A clinical high-resolution acquisition was employed (15 min/bed; OSEM 4 iterations, 28 subsets, 3mm post-filter, 2 mm^2 pixels). Results: Analysis of the apparent values in the 15 features yielded contrast recovery curves demonstrating reduced contrast for smaller features. A threshold of detectability was defined as having a value greater than the mean plus two standard deviations of the background values, reflective of a >95% likelihood of seeing a feature assuming normally distributed noise. This threshold confirmed the visual assessment of the images showing that features with contrast of 5:1 and a size less than 2.7 mm across were not detectable. The maximum values of the 15 features sampled the 2D space of different contrast and sizes; This 2D surface was interpolated to create isocontour lines of shared maximum SUV values, permitting estimation of the detectability of different types of features for a tested scanner, protocol, and count density. Furthermore, this single scan provides the contrast recovery curves that could be used for partial volume correction of features with known size. Conclusions: We present a new phantom design to characterize the partial volume errors of small features in PET imaging. This phantom and subsequent analysis provides an estimate of the necessary contrast and size to permit detection of these features and the associated contrast recovery curves. Acknowledgment: This work was supported through NIH grant R01 HL135557. ER -