Assessment of myocardial perfusion by dynamic O-15-labeled water PET imaging: validation of a new fast factor analysis

Abstract

Background: Factor analysis (FA) is an established method for separating myocardium from blood pool by use of oxygen 15-labeled water and positron emission tomography for analyzing myocardial blood flow (MBF). Conventional FA methods generating images from sinograms (sinoFA) are time-consuming, whereas FA can be performed on the reconstructed images (reconFA) in a fraction of time. We validated the MBF values obtained by reconFA versus sinoFA.

Methods and results: In 23 volunteers (mean age, 26.6 +/- 3.4 years) MBF was calculated from sinoFA and reconFA and blindly reanalyzed 1 month later by the same observer. Intraobserver agreement and reconFA-versus-sinoFA agreement were assessed according to Bland and Altman (BA). Reproducibility proved excellent for global sinoFA (r = 0.968; P < .001; BA limits, -0.617 to 0.676 mL x min(-1) x g(-1)) and slightly superior for reconFA (r = 0.979; P < .001; BA limits, -0.538 to 0.558 mL x min(-1) x g(-1)), with wider limits of agreement for segmental MBF from sinoFA (r = 0.777; P < .001; BA limits, -1.676 to 1.656 mL x min(-1) x g(-1)) and reconFA (r = 0.844; P < .001; BA limits, -1.999 to 1.992 mL x min(-1) x g(-1)). In addition, sinoFA and reconFA showed excellent correlation (r = 0.975, P < .001) and agreement (BA limits, -0.528 to 0.648 mL x min(-1) x g(-1)) for global and segmental values (r = 0.955; P < .001; BA limits, -1.371 to 1.491 mL x min(-1) x g(-1)).

Conclusions: Use of reconFA allows rapid and reliable quantitative MBF assessment with O-15-labeled water.