Abstract
435
Objectives The aim of this study is to optimize the design of a time-of-flight (TOF)-depth-of-interaction (DOI) detector, based on stacked crystal layers, to be used in a long axial field-of-view, whole-body PET scanner. A combination of axial and radial parallax error is encountered when imaging with the proposed geometry, exacerbating the need for a DOI correction. Here we present an optimization framework, based on Monte Carlo simulations and select experimental measurements, to guide the choice of layer thickness in a multi-layer DOI detector with TOF capability.
Methods Coincidence timing resolution (CTR) was characterized experimentally using stacks of scintillator pixels (both LaBr3 and LYSO), creating multi-layer DOI designs having layers of different thicknesses. DOI discrimination is performed using signal rise time characteristics. Timing was measured in each layer, and a block CTR estimated using photon detection efficiencies for each layer. Imaging simulations were performed to assess the effect of DOI on image quality. The simulated scanner had an axial length of 72cm and a diameter of 85cm, and used 4 mm wide crystals with DOI having adjustable layer thickness (total thickness fixed at 20mm). Two phantoms were used, measuring lesion contrast and spatial resolution.
Results DOI by rise time discrimination allows for identical scintillators to be used in both detector layers. When using layers of different thicknesses CTR can vary significantly between layers, though after accounting for the detection efficiencies of each layer, which is also dependent upon the thickness, the layer depth does not greatly change the average CTR for our DOI design. Imaging simulations show that DOI improves both spatial resolution and image contrast.
Conclusions In long axial FOV scanners, a DOI detector can mitigate the large parallax error and improve image quality. A methodology is created, based on TOF capability, and image resolution and contrast, to optimize the layer depth of our dual-layer DOI detector design and can also be applied to other DOI concepts.
Research Support NIH R01 CA113941, R21 EB016205, and R01 EB009056.