Impact of low-dose SPECT imaging on normal databases and myocardial perfusion scores

Purpose: We have recently demonstrated that iterative reconstruction algorithms with resolution recovery require the adoption of specific normal databases (NDBs) for perfusion SPECT quantification. This work was aimed at investigating the impact of patient low-dose imaging on NDBs and percent summed rest (SR%) and stress (SS%) scores. Methods: Assuming that count statistics of shorter acquisition time may simulate that of lower patient dose, three simultaneous scans were acquired (BrightView, Philips) with different acquisition-time/projection: 30, 15 and 8 s (from 100% to 25% of the reference). Fifty-two normal patients with low likelihood of coronary artery disease were enrolled and three homemade NDBs were then generated and compared (Astonish™ algorithm with default parameters): 100%-HM-NDBs, 50%-HM-NDBs and 25%-HM-NDBs. SR% and SS% were subsequently calculated for another group of 38 patients (normal/abnormal = 5/33). SR% and SS% values of 100%-HM-NDBs were compared with those obtained with the NDBs available on the workstation. Moreover, the impact of the study count statistics on perfusion scores was evaluated using the count-specific NDBs. Results: Significantly higher standard-deviation values were found for 25%-HM-NDBs compared to the other HM-NDBs (p < 0.02). Significantly higher SS% were also found for the 100%-HM-NDBs compared to the workstation NDBs (95%CI: 0.15–2.11%). Moreover, a post-hoc test showed significantly lower SR% and SS% for 25%-count statistics compared to 100%-HM-NDBs (p < 0.03). Conclusions: NDBs and perfusion scores depend significantly on study count-statistics. A 50% reduction in patient dose is ultimately the limit for Astonish™ (with the default parameters) in order to prevent a significant variation in myocardial perfusion quantification.