Impact of injected dose and acquisition time on a normal database by use of 3D-SSP in SPECT images: quantitative simulation studies

Abstract

The present study aimed to validate the accuracy of normal databases (NDBs) with respect to variable injected doses and acquisition times by use of three-dimensional stereotactic surface projections (3D-SSP) in N-isopropyl-p-[123I]-iodoamphetamine (I-123-IMP) brain perfusion images. We constructed NDBs based on brain SPECT images obtained from 29 healthy volunteers. Each NDB was rebuilt under simulated unique conditions by use of dynamic acquisition datasets and comprised injected doses (222, 167, and 111 MBq) and acquisition times (30, 20, and 15 min). We selected seven of 29 datasets derived from the volunteers to simulate patients' data (PD). The simulated PD were designed to include regions of hypoperfusion. The study comprised protocol A (same conditions for PD and NDB) and protocol B (mismatched conditions for PD and NDB). We used 3D-SSP to compare with the Z score and detection error. The average Z scores were decreased significantly in protocol A [PD (High)-NDB (High) vs. PD (Low)-NDB (Low); PD (30 m)-NDB (30 m) vs. PD (15 m)-NDB (15 m) and PD (20 m)-NDB (20 m)].The average Z scores of PD (High) and PD (Medium) with NDB (High) did not differ significantly in protocol B, whereas all others were decreased significantly. The error of detection increased 6.65 % (protocol A) and 32.05 % (protocol B). The Z scores were specific to the injected dose and acquisition time used in 3D-SSP studies, and the calculated Z scores were affected by mismatched injected doses and acquisition times between PD and selected NDBs.