SPECT blood pool phase analysis can accurately and reproducibly quantify mechanical dyssynchrony

Abstract

ObjectivesPhase analysis of SPECT blood pool imaging has the potential to assess mechanical dyssynchrony (MD). However, wall motion of the left ventricle (LV) from SPECT images can be based on either time-activity or time-distance curves. In this paper, these two techniques were compared using receiver-operator characteristics (ROC) analysis at detecting MD patients from a population of normal subjects. MethodsSPECT phase analysis was performed on 48 normal subjects (LVEF > 55%, normal wall motion, QRS < 120 ms), and 55 MD patients (LVEF < 35%, QRS > 120 ms). ROC analysis was individually performed on each of three phase parameters (phase standard deviation, synchrony, and entropy) for each LV wall motion technique. ROC area differences were assessed using the Student t-test. Intra- and inter-observer reproducibilities were investigated using regression analysis. ResultsTime-activity-based phase analysis produced excellent ROC areas of .93 or better for all three phase parameters. The time-distance techniques produced significantly (P < .05) lower ROC areas in the range of .53-.76. Time-activity-based phase analysis had excellent intra- and inter-observer reproducibility with correlation coefficients >.96, compared to values of ~.85 for the time-distance methods. ConclusionSPECT time-activity-based phase analysis had excellent sensitivity and specificity at detecting MD patients with very high intra- and inter-observer reproducibility.