Optimized Methodology for Reference Region and Image-Derived Input Function Kinetic Modeling in Preclinical PET

Abstract

PET imaging of small animals is often used for assessing biodistribution of a novel radioligand and pharmacology in small animal models of disease. PET acquisition and processing settings may affect reference region or image-derived input function (IDIF) kinetic modeling estimates. We examined four different factors in comparing quantitative results: 1) effect of reconstruction algorithm, 2) number of MAP iterations, 3) strength of the MAP prior, and 4) Attenuation and scatter. The effect of these parameters has not been explored for small-animal reference region and IDIF kinetic modeling approaches. Dynamic PET/CT scans were performed in 3 species with 3 different tracers: house sparrows with 11Craclopride, rats with 18FAS2471907 (11bHSD1) and mice with 11CUCB-J (SV2A). FBP yielded lower kinetic modeling estimates compared to 3D-OSEM-MAP reconstructions, in sparrow and rat studies. Target resolutions (MAP prior strength) of 1.5 and 3.0mm demonstrated reduced VT in rats but only 3.0mm reduced BPND in sparrows. Therefore, use of the highest target resolution (0.8mm) is warranted. We demonstrated using kinetic modeling that forgoing CT-based attenuation and scatter correction may be appropriate to improve animal throughput when using short-lived radioisotopes in sparrows and mice. This work provides recommendations and a framework for future optimization of kinetic modeling for preclinical PET methodology with novel radioligands.