Quantitative myocardial mouse heart imaging with positron emission tomography (PET) is compromised when the PET system has a resolution on the order of the left ventricular (LV) wall thickness. In this study a partial volume correction (PVC) strategy is employed to compensate for the partial volume losses experienced in PET images from the Inveon small animal PET scanner. A 1D algorithm extracts the edges of the heart wall and activity information outside the wall. This information is used to build a 2D image and a PVC model is developed to correct the activity in the LV. A dynamic mouse heart LV phantom, with variable wall thickness and 8 gates, was created digitally and physically so that the algorithm could be tested on simulated and real measured image data. An effective 2D Gaussian point spread function, with a FWHM of 1.3 mm, was derived from the measured PET images and used in the PVC models. The simulated and measured phantom images underestimated the relative activity by approximately 30–40% before PVC, but this was improved to an overestimation of 10% after PVC. The coefficient of variation was improved from 20–30% to 5–10% with PVC demonstrating that the algorithm has the ability to increase accuracy and restore homogenous activity distributions in 2D. This algorithm was developed for mouse imaging, but it could be applied to any other species where myocardial wall thickness is on the order of the PET system's spatial resolution.

Additional Metadata
Persistent URL dx.doi.org/10.1118/1.3476183
Journal Medical Physics
Citation
Dumouchel, T., & Dekemp, R. (2010). Sci—Fri AM: Imaging — 04: Experimental Validation of a Two Dimensional Partial Volume Correction Strategy for PET Imaging in Mice with Simulated and Measured Data. In Medical Physics (Vol. 37). doi:10.1118/1.3476183