PSMA PET-CT: The effects of decreasing scan times on image quality.

2000 Objectives: Prostate-specific membrane antigen (PSMA)-targeted radiotracers used in PET-CT imaging such as 18F-DCFPyL, have shown higher sensitivity compared to conventional imaging for unveiling disease foci in patients with prostate cancer. The purpose of this study was to evaluate the effects of decreasing PET scan times on image quality by using three metrics: contrast-to-noise ratio (CNR) in the foci, signal-to-noise ratio (SNR) in the liver, and coefficients of variation (CV) of image noise. Additionally, we evaluated the clinical impact of decreased scan time on the Prostate Cancer Molecular Imaging Standardized Evaluation (PROMISE) PSMA score for foci with the lowest and highest uptake values. Methods: Patients were imaged first with PET-MR at 2 hr and then PET-CT approximately 3.5 hr after 18F-DCFPyL injection. Because of the longer uptake time for the PET-CT, whole body PET scans were acquired for 5 min/bed in list mode. List mode data were then rebinned into scan times of 4, 3, 2.5, and 2 min. Patients with two or more reportedly-positive sites were included in this study (n=11). SUV values (max, mean, and peak) were measured for liver, left ventricular cavity for blood pool (BP), and parotids. Liver and BP SUV values were determined within fixed spherical volume-of-interest (VOI) templates (30 and 20 mm diameter) manually placed on PET images using the CT for guidance. Similarly, VOIs were placed on each parotid gland and averaged to derive the parotids SUVmax. Additionally, in each patient the two PSMA-positive foci with low and high uptake were quantified. For each of these, a bounding box (BB) VOI was placed and SUVmean was measured using a 42% threshold of the max value within the BB. To measure SUVpeak, the same BB was used and a 1 cc sphere was centered over the max value. All VOI placements were performed on the 5 min scan and using an independent third-party workstation, the other scan times were coregistered so the same VOIs were applied to all scans. CNR was calculated as (SUVmax,foci - SUVmean,BP)/SUVsd,BP for the foci with lowest SUVmax only. SNR was calculated as SUVmean/sd. CV was calculated as (SUVsd/SUVmean) x 100% for liver, BP, and both foci. PSMA scores were determined for both low and high foci at all scan times by comparing the SUVmean to BP, liver, and parotid as per the PROMISE PSMA scoring system (Eiber et al. J Nucl Med 2018). Results: All SUV values for the low foci (see table) and high foci varied only minimally across the 5 scan times for all patients. In liver, SUVmax varied the most, increasing with decreasing scan times. As expected, CNR low foci data indicate worsening image quality with shorter scan times. SNR for liver shows the same trend: as scan time decreases, the SUVsd in the image increases, thus reducing the SNR. The overall averages of CV for liver, BP, high and low foci (max, mean and peak) ranged from 0.7 to 8.1% with only 4 out of 132 individual values greater than 10%, thus indicating low image noise. PSMA scores changed only in one patient for the high SUV foci from a score of 2 (PSMA expression equal to or above liver and lower than parotid) at both the 2 and 2.5 min scan times, to scores of 1 (equal to or above blood pool and lower than liver) for the 3, 4, and 5 min scan times. Conclusions: Although CNR decreases with shorter scan time, even at 2 min the average CNRmax for all patients was 16 and the lowest individual value was 6, indicating good image quality. SNR in liver was 10 at 3 min, which meets the criteria of good image quality according to recommendations of Fukukita et al. (Ann Nucl Med 2014). Image noise as measured by CV was below the suggested limit of 10% of Akamatsu et al. (J Nucl Med Technol 2015). The PSMA score decreased in only one patient on the two shortest scan times (2 and 2.5 min), warranting further studies to determine the incidence and clinical implications of this observation. In summary, based on these results, our current scan time of 4 min/bed could be safely reduced to 3 min, thereby reducing the total scan time by 8 min from 32 to 24 min.