Influence of Depth of Interaction on Spatial Resolution and Image Quality for the HRRT

Abstract

The high resolution research tomograph (HRRT) is an octagonal design PET camera with 119,808 crystals arranged in a dual layer to determine the depth of interaction (DOI) and compensate for the parallax effect. The DOI discrimination is based on the identification of the layer in which the gamma interaction occurred using pulse shape discrimination. However the observed fractional crystal efficiency is count rate dependent, thus affecting the accuracy of the pulse shape discrimination. In this study we investigated the impact of the mismatch between the emission and the normalization scan count rate on image uniformity using phantom data when DOI correction was applied and when it was switched off. Count rate mismatch was found to manifest itself in form of streaking artifacts and high frequency non-uniformities with a star shape pattern in Fourier space. It was found to be enhanced when DOI correction was applied. In realistic scanning conditions assessed with non-human primate data the effect of count rate mismatch was found to be nearly negligible with DOI correction present or absent. Since DOI corrected data proved to be more sensitive to an emission/normalization count rate mismatch, the impact of DOI on resolution and biological measure obtained in realistic scanning conditions was further evaluated. With DOI determination, spatial resolution was improved by up to 7% in the outer part of the FoV where it was measured to be 2.9 /spl plusmn/ 0.2 mm (SPAN 3) and 3.3 /spl plusmn/ 0.2 mm (SPAN 9) and the biological parameters (binding potentials) extracted from the non-human primate study were improved by up to 5%. In summary this study shows a greater sensitivity to emission/normalization count rate mismatch in phantom studies when DOI correction is present. However much less sensitivity is observed in realistic data, while the resolution uniformity advantage due to DOI determination is still noticeable, not only in resolution measurement but also in the accuracy of the biological measures extracted from realistic scanning protocols.