Regional left ventricular function from short-duration SPECT gated blood pool studies

SPECT gated blood-pool (GBP) scans can provide accurate 3D assessments of resting regional left-ventricular (LV) function. However, due to limited scan times for stress studies, SPECT GBP images are often too noisy for standard visual analysis. In this investigation, we examined whether quantitative regional analysis could be performed despite this increased noise. A count-based scheme for measuring the regional ejection fraction (REF) and relative regional phase (RRP) was investigated. Constructing a rest normal database (NLDB) from the 30-minute SPECT GBPs of 10 volunteers, the mean and standard deviation (σ_{(rest NLDB)}) of these parameters was calculated for each of 12 3D regions. Adding statistical noise, the expected NLDB variability at stress (σ_{(stress NLDB)}) from 3-minute scans was estimated for each parameter. The variability due to increased noise (σ_noise) resulted in an average increase from σ_{(rest NLDB)} to σ_{(stress NLDB)} of 19% in REF, and 45% in RRP. Analyzing rest scans from two patients with regional defects, the regions were classified twice; first based on a criterion of ±2σ_{(rest NLDB)} about the average NLDB value for that region, then again using ±2σ_{(stress NLDB)}. Of the 19 abnormal regions identified at rest, 18 were still detected with the ±2σ_{(stress NLDB)} criterion. This suggests that quantitative regional analysis may be relatively insensitive to the increased noise from a 3-minute scan, making stress SPECT GBP studies potentially feasible