Performance comparison of three commercially available PET systems : SIGNA PET/MR, Discovery IQ and Discovery MI

• Objectives: The NEMA performance measurement standard (NEMA NU 2-2012) for PET scanners provides guidelines on how to assess the performance of Positron Emission Tomography (PET). Three different state of the art PET systems were installed at Uppsala University Hospital between years 2014-2016 and independent NEMA standard tests were performed. The aim of this study was to compare system performance of the three scanners.Methods: Three commercially available scanners from GE-healthcare (SIGNA PET/MR; Discovery IQ PET/CT; Discovery MI PET/CT) were evaluated. The SIGNA and MI systems are based on LYSO crystals and digital SiPMs, whereas the IQ uses BGO crystals and regular PMTs. Spatial resolution, sensitivity, count rate statistics, count rate accuracy and image quality were assessed according to the NEMA NU 2-2012 standards. In addition to the NEMA standard test, recovery was assessed for different reconstructions using the NEMA image quality phantom at a contrast of 4:1 in all spheres, and a triple line insert phantom. These tests were performed on all three scanners in a single session, avoiding differences due to variability in phantom preparation.Results: Full width of half maximum (FWHM) of the spatial resolution (radial/tangential/axial) reconstructed with filtered back projection (FBP) at 1,10 and 20 cm from the centre of FOV is illustrated in figure 1A for each system. The average sensitivity, Peak NECR, scatter fraction and count rate accuracy of each system is presented in table 1. The average image contrast recovery coefficients of SIGNA, IQ and MI varied between 45, 40 and 56 % (10 mm sphere) to 74, 72 and 84 % (22 mm sphere) respectively. The average image contrast recovery coefficients is presented in figure 1B. The lung error for SIGNA, IQ and MI were 2.7, 18 and 5.2 % respectively. Using reconstruction settings recommended for clinical use (Signa: TOF-OSEM, 2 iterations/28 subsets, 5 mm post-filter; IQ: OSEM, 4/12, 4 mm; MI: TOF-OSEM, 3/16, 5 mm, all with resolution recovery) recovery based on a volume of interest over whole spheres varied between 50, 38 and 51 % (10 mm sphere) to 86, 83 and 87 % (22 mm sphere), respectively. In addition to the recommended settings for clinical use, Q.Clear (Block-sequential regularized expectation maximization (BSREM) with PSF modeling) reconstructions with beta values ranging from 100 to 500 with step of 200 were reconstructed. The volume recovery of each system for varying reconstructions is presented in Figure 1C. The mean radial/tangential/central spatial resolution of SIGNA, IQ and MI using the triple line insert phantom when using the recommended standard reconstructions and Q.Clear reconstruction is presented in Figure 1D.Conclusion: As expected, the two ToF systems based on LYSO crystals coupled to digital SiPMs (SIGNA and MI), resulted in an overall better resolution, image quality, NECR and volume recovery than the non-TOF BGO coupled to non-digital detector system (IQ). The image quality and spatial resolution improved when Q.Clear reconstruction was used. The sensitivity was higher in SIGNA than in MI and IQ due to a 25 cm axial FOV in SIGNA, compared to 20 cm for MI and IQ. In conclusion, the new SiPM-based PET detector systems provide a considerable enhancement in system performance.

Ämnesord

MEDICIN OCH HÄLSOVETENSKAP  -- Klinisk medicin -- Radiologi och bildbehandling (hsv//swe)

MEDICAL AND HEALTH SCIENCES  -- Clinical Medicine -- Radiology, Nuclear Medicine and Medical Imaging (hsv//eng)

Publikations- och innehållstyp

vet (ämneskategori)

art (ämneskategori)