Abstract
252184
Introduction: High spatial resolution and high sensitivity are required for brain PET to visualize small structures in the brain. Spatial resolution of PET depends on various factors such as a scintillation crystal pitch, the photon non-collinearity effect (angular deviation), the positron range, and the parallax error (depth-of-interaction (DOI) encoding error). A brain-sized compact geometry, fine scintillator pitch, and DOI measurement capability are key specifications to achieve high spatial resolution. Recently, we have developed a novel PET detector with the crosshair light-sharing (CLS) configuration that offers both time-of-flight (TOF) and DOI measurement capabilities. The scintillator pitch of the CLS detector is 1.6 mm, which is finer than other commercial brain PET systems. The purpose of this study is showing a proof-of-concept of the CLS detector by prototyping a one-ring TOF-DOI brain PET.
Methods: The fast lutetium gadolinium oxyorthosilicates (LGSO) (OXIDE, Japan) scintillators each sized at 1.45×1.45×15 mm3 were attached to a silicon photomultiplier (SiPM) (Hamamatsu Photonics, K.K., Japan) array of 3.0×3.0 mm2 pixels. The detector block consisted of a 14×14 scintillation crystal array and an 8×8 SiPM array. Two neighboring scintillators were optically glued to allow light to pass through, and other parts were covered by enhanced specular reflectors (ESRs) (3M, St. Paul, MN, USA). The combination of the optical glue and the reflectors allows scintillation light to loop within these crystals and to spread mainly among two neighboring SiPMs under a pair of crystals. DOI information was calculated from the output ratio of these two SiPMs. Thirty-six detector blocks were circularly arranged with the crystal-to-crystal diameter of 30 cm. The axial field-of-view was 2.6 cm. We used the PETsys TOFPET2 ASIC for data acquisition. Energy, timing and DOI (3-segments) calibration and normalization were performed with measuring an 18F annular distribution phantom. We evaluated the energy resolution, TOF resolution and spatial resolution, followed by imaging demonstration of the Hoffman 3D brain phantom. Only the four acrylic plates and the outer case were used so that the radioactivity was distributed within the FOV. Image reconstruction was performed with the ordered-subset expectation-maximization algorithm. To investigate the effect of the reconstruction with TOF and DOI, non-TOF and non-DOI PET data were made by post-processing.
Results: The energy resolution at 511 keV was 11.6% and the TOF resolution was 297 ps. For the spatial resolution measurement, one-millimeter rods were resolved even at the 10 cm offset position. The rod separation capability was clearly improved with DOI information. In the Hoffman brain phantom images, the gray and white matter areas were visualized with high contrast. Both TOF and DOI information effectively improved the contrast. The 0.8-mm-thick radioactivity distributions could be identified in the coronal image.
Conclusions: We developed a one-ring prototype of a TOF-DOI brain PET system with our original CLS detectors. One-millimeter rod separation capability, 0.8-mm-thick structure visibility, and 297 ps TOF resolution were demonstrated experimentally.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.