Citation:
Physics in Medicine and Biology, (2013), 58(7), 2059-2072.
ISSN:
0031-9155 (Print) 1361-6560 (Online)
DOI:
10.1088/0031-9155/58/7/2059
Sponsor:
This work was partially funded by AMIT project (CEN-20101014) from the CDTICENIT program, CIBERsam (CB07/09/0031), projects TEC2010-21619-C04-01 and TEC2011-28972-C02-01 from Spanish Ministerio de Ciencia e Innovación, Spanish Government (ENTEPRASE Grant, PSE-300000-2009-5), PRECISION grant IPT-300000- 2010-3, CPAN (CSD-2007-00042@Ingenio2010), MEC (FPA2010-17142) and ARTEMIS program (S2009/DPI-1802) from Spanish Comunidad de Madrid and EU-ERDF program. Part of the calculations of this work were performed in the "Clúster de Cálculo de Alta Capacidad para Técnicas Físicas" funded in part by UCM and in part by UE with European regional funds. This is a contribution from the Moncloa Campus of International Excellence.
Pile-up and dead-time are two main causes of nonlinearity in the response of a PET scanner as a function of activity in the field of view (FOV). For a given scanner and acquisition system, pile-up effects depend on the material and size of the object being imaPile-up and dead-time are two main causes of nonlinearity in the response of a PET scanner as a function of activity in the field of view (FOV). For a given scanner and acquisition system, pile-up effects depend on the material and size of the object being imaged and on the distribution of activity inside and outside the FOV, because these factors change the singles-to-coincidences ratio (SCR). Thus, it is difficult to devise an accurate correction that would be valid for any acquisition. In this work, we demonstrate a linear relationship between SCR and effective dead-time, which measures the effects of both dead-time (losses) and pile-up (gains and losses). This relationship allows us to propose a simple method to accurately estimate dead-time and pile-up corrections using only two calibration acquisitions with, respectively, a high and low SCR. The method has been tested with simulations and experimental data for two different scanner geometries: a scanner with large area detectors and no pile-up rejection, and a scanner composed of two full rings of smaller detectors. Our results show that the SCR correction method is accurate within 7%, even for high activities in the FOV, and avoids the bias of the standard single-parameter method.[+][-]