PT - JOURNAL ARTICLE AU - Hua Qian AU - Ravindra Manjeshwar AU - Chengeng Zeng AU - Paul Kinahan AU - Lawrence MacDonald TI - Tradeoffs in Detector Design for a Rectangular Breast PET Scanner DP - 2015 May 01 TA - Journal of Nuclear Medicine PG - 1856--1856 VI - 56 IP - supplement 3 4099 - http://jnm.snmjournals.org/content/56/supplement_3/1856.short 4100 - http://jnm.snmjournals.org/content/56/supplement_3/1856.full SO - J Nucl Med2015 May 01; 56 AB - 1856 Objectives We are developing a rectangular scanner for dedicated breast PET imaging to be mounted on conventional mammography equipment (PET/X) [1]. A critical design parameter is detector thickness, as thinner detectors reduce cost and improve resolution, but also decrease sensitivity [2]. We conducted a simulation study to evaluate bias/variance as a function of detector thickness and lesion size. Our choice of detector thickness depends on the contrast recovery coefficient (CRC) accuracy vs. precision trade-off.Methods We simulated PET/X using SimSET. Detector panels were made of 2x2 mm LYSO crystals and we evaluated 5, 10 and 15 mm crystal thicknesses. Spherical lesions of 2-10 mm and 5:1 source to background ratio were inserted into a 70 mm thick breast phantom. Two hundred million decays were simulated, which correspond to a 2-minute breast scan 2 hours after a 10 mCi FDG injection. Fully 3D image reconstruction algorithms with sensitivity and attenuation corrections were developed for the rectangular scanner. The geometrical sensitivity variations for the parallel and orthogonal detector panels were incorporated into the distance-driven projectors [3]. The detector sensitivity variation due to the difference in incident angles was calculated as a map of detector normalization. CRC accuracy and precision were computed as figures of merit, as well as lesion signal to background noise ratio (SNR), which is linearly correlated to ensemble SNR [4].Results Depending on lesion size, CRC values decreased by 69-82% and 43-71%, but the lesion SNR increased by 48-62% and 71-93% as the crystal thickness was increased from 5 to 10 and 15 mm respectively. In addition, for the range of simulated lesion sizes, CRC showed a stronger dependence on lesion size compared to SNR.Conclusions Crystals thickness of 10 mm appears to provide a suitable compromise between CRC accuracy and precision needed for using PET to monitor response to breast cancer therapy.Research Support This work was supported by the National Institute of Health under Grant R01-CA163498.