Methods to compensate the time walk errors in timing measurements for PET detectors

We investigate the effects of the signal walk-in timing measurements and the methods to compensate for the error when leading-edge discriminator (LED) triggering was used. Three pairs of crystals (LaBr3 cubes, lutetium-yttrium oxyorthosilicate (LYSO) cubes, and LYSO rectangular prisms) were tested for coincidence timing resolution (CTR) to evaluate four compensation methods: 1) Linear 1-D; 2) Logarithm 1-D; 3) Polynomial 2-D; and 4) artificial neural network (ANN) 2-D. The experimental results show that: 1) the signal walk causes a logarithmic relationship between energy and measured time; 2) the signal walk dramatically affects the timing measurement at a wide energy window. The measured FWHM CTR decreased by 86.2%, from 670 ps to 92.7 ps, when the lower cut of the energy window was used; and 3) the Logarithm 1-D compensation can improve the timing performance. The FWHM CTRs decreased by 69.9% (from 326.8 ps to 101.3 ps) with an energy window of [300 keV–550 keV] and by 12.5% (from 97.2 ps to 85.1 ps) with an energy window of [410 keV–550 keV] for LaBr3 cube crystals. The results show that the Logarithm 1-D, Polynomial 2-D, and ANN 2-D compensations work well, achieving at least a 10% improvement in timing performance.