A pre-clinical PET scanner based on a single piece of annular scintillator: a simulation study

Objectives One challenge of pre-clinical PET imaging is the amount of scan time required to produce high quality, high-resolution images. This issue is important because many pre-clinical studies require the rapid scanning of numerous animals. To create a very high sensitivity and high resolution pre-clinical PET scanner, one based on a single scintillator crystal formed into an annulus is proposed. The objective of this study was to evaluate, via simulation, the potential performance of this system.

Methods The simulated scanner consisted of a 72.7mm long cylindrical piece of LYSO crystal with a diameter of 75 mm. The cylinder was formed into an annulus by machining a hole (diameter of 50mm) in the center of the scintillator. Twelve flat surfaces were formed on the outer surface of the annulus to accommodate light detection devices. Specifically, an array of 5 by 17 discrete SiPMs with a size of 3mmx3mm (pitch of 4.1mm, similar to the SensL C-series SiPMs) was attached to each facet. Evaluation of scanner performance was carried out in three steps. First, GATE simulation software was used to obtain a list mode distribution of detected 511keV photons. Next, a trial GATE data set was used in conjunction with the DETECT2000 optical simulation software to explore light transport in the detector. From these results, the three-dimensional positions of the simulated positron annihilations in the scintillator were calculated from the detected optical photons distribution detected at the surfaces of the SiPMs (depth of the events were estimated by the shape of the light distribution). Calculated positions were compared to original emission points to determine spatial accuracy of the event reconstruction. Finally, the data were histogrammed into 100 radial and 180 angular bins and reconstructed using the SSRB-FBP algorithm. Images from a point source and the brain section of the MOBY phantom were created. Data from the point source were used to measure spatial resolution and detection sensitivity.

Results Event positioning was accurate to within 1.9mm in the tangential and radial directions. Interaction depth resolution was ~4.4mm. Detection sensitivity at the center of the scanner was 10.3% which is more than twice of any pre-clinical scanner of comparable axial size and diameter. Spatial resolution at the center of the scanner was 1.04mm FWHM. Finally, images of the MOBY mouse brain phantom resulted in good results for FBP reconstruction.

Conclusions The results from this initial study demonstrated the promise of a solid, annular pre-clinical PET scanner.

• Download figure
• Open in new tab
• Download powerpoint