Validation of Gated Blood-Pool SPECT Regional Left Ventricular Function Measurements

Abstract

Blood-pool (BP)-gated SPECT should be able to detect regional left ventricular (LV) dysfunction, as the modality is fully 3-dimensional and capable of resolving all cardiac chambers. This study investigates the hypothesis that LV segments that have abnormal regional wall motion (WM) on a cardiac MRI scan also have abnormal BP regional ejection fraction (EF) as computed by fully automated quantitation (AQ) of BP data. A total of 34 patients evaluated for coronary artery disease who underwent visual assessment of WM by review of BP cines and cardiac MRI evaluations were included in this retrospective investigation. Cardiac MRI values for these patients were compared with an institutional database of cardiac MRI values for an age-matched cohort of 10 healthy volunteers. An AQ algorithm segmented the LV BPs on the BP tomograms and subdivided volumes into 17 subregions. Count-versus-time curves were fit to third-order Fourier series for each LV subvolume to compute regional EFs. For cardiac MRI data, endocardial and epicardial drawings were performed manually for 60 degrees samples of 11-13 short-axis tomograms spanning the entire heart, from which regional WM values were computed and rebinned into 17 conventional LV segments. Global EF ranged from 12% to 75% on AQ and from 14% to 75% on cardiac MRI (Pearson correlation coefficient=0.95, P<0.0001). Differences were not significant between BP AQ and cardiac MRI in identifying the 12 patients with a global EF less than 35% (McNemar difference, 3%; P=1.0) and the 19 patients with an EF less than 50% (difference, 3%; P=1.0). BP AQ was more accurate than was visual assessment for identifying LV segments with abnormal cardiac MRI WM (receiver-operating-characteristic areas, 88% vs. 80%, P<0.0001) and was more accurate for the left circumflex than for the left anterior descending coronary artery territories (95% vs. 86%, P=0.01). Differences were not significant between BP AQ and cardiac MRI WM for discriminating normal from abnormal LV segments (McNemar difference, 3.2%; P=0.14). AQ BP-gated SPECT assessment of regional and global LV WM agrees with independent cardiac MRI calculations and is superior to visual analysis for detecting regional WM abnormalities.