RT Journal Article SR Electronic T1 3D Model-based partial volume correction for cardiac PET imaging in mice JF Journal of Nuclear Medicine JO J Nucl Med FD Society of Nuclear Medicine SP 204 OP 204 VO 52 IS supplement 1 A1 Tyler Dumouchel A1 Robert deKemp YR 2011 UL http://jnm.snmjournals.org/content/52/supplement_1/204.abstract AB 204 Objectives Quantitative cardiac PET imaging is limited by spatial resolution. In this study a 3D partial volume correction (PVC) algorithm is developed for mouse LV myocardial imaging. Methods A 3D model of the ECG-gated cardiac geometry and activity was estimated and convolved with the scanner PSF. Regional parameters of the model (LV myocardium, blood and background activity, and epi/endocardial wall borders) were varied to fit measured image data. ECG gated cardiac mouse PET images, with and without noise, were simulated using the MOBY phantom convolved with a Gaussian PSF. Images were simulated with axial pixel sizes of 0.2 and 0.8 mm and resolutions of 1.25 and 1.8 mm. Images were generated with the LV oriented axially and at 20 degrees off-axis. The algorithm was evaluated in six healthy mice injected with 18F-FDG, scanned with the Inveon and reconstructed with 8 ECG gates. Results The PVC reduced bias in LV myocardial activity from 35% to within 5% and decreased the COV from 11-12% to 6-9% on the simulated images, demonstrating the ability to simultaneously improve recovery and homogeneity. With the heart oriented on-axis, an axial resolution of 1.8 mm did not affect the corrected activity, however an axial pixel size of 0.8 mm caused a 3% underestimation (p <0.01) of activity, likely due to undersampling at the apex. Off-axis orientation of the heart did not affect corrected activity. Typical noise levels did not affect bias in corrected activity with the larger axial pixel size or resolution, although it did have an effect (p<0.05) when the heart was oriented off-axis. The PVC increased recovery in the gated mouse hearts by 30% and the COV was improved by 3% on average, demonstrating that the PVC performed comparably to the simulations. Conclusions PVC is necessary to restore quantitative accuracy in cardiac imaging. The 3D PVC algorithm explored in this study appears to be a feasible solution to improve quantitative accuracy in mouse heart imaging with PET. Research Support NSERC, CIH