Reproducibility and accuracy of gated SPECT for determination of left ventricular volumes and ejection fraction: experimental validation using MRI.

UNLABELLED: Quantitative gated SPECT (QGS) has been used for computation of left ventricular volumes and ejection fraction. This study evaluated, first, the effect of injected dose, time of imaging, and background activity on the reproducibility of QGS and, second, the accuracy of QGS, compared with cine MRI, for determining left ventricular volumes and ejection fractions in dogs with and without perfusion defects. METHODS: Sixteen dogs were subjected to either chronic occlusion of the circumflex artery (group I, no perfusion defect) or acute occlusion of the anterior descending coronary artery (group II, perfusion defect). Both groups underwent serial MRI and SPECT. RESULTS: ( QGS was very reproducible using the automated program (r = 0.99997). Correlation between left ventricular ejection fraction (LVEF) at 15 and 45 min was poor after the low-dose injection (r = 0.54; SE = 9%) and only fair after the high-dose injection (r = 0.77; SE = 5%). Correlation was poor in the presence of significant background activity (r = 0.36; SE = 12%). Correlation between QGS left ventricular volumes and MRI was good for group I (end-diastolic volume, r = 0.86; end-systolic volume, r = 0.81) and only fair for group II (end-diastolic volume, r = 0.66; end-systolic volume, r = 0.69). The overall LVEF correlation between QGS and MRI was poor (r = 0.51). QGS LVEF (mean +/- SD, 42% +/- 3%) overestimated MRI LVEF (29% +/- 2%). CONCLUSION: QGS provides a highly reproducible estimate of LVEF. However, QGS is affected by changes in background activity, time of imaging, and injected dose. In the presence of perfusion defects, QGS overestimated volume relative to MRI. The correlation between QGS- and MRI-derived LVEF was poor in this canine model.

Quantification of regional myocardial wall thickening on electrocardiogram-gated SPECT imaging.

BACKGROUND: Current assessment of regional left ventricular function with electrocardiogram (ECG)-gated single photon emission computed tomography (SPECT) imaging is generally performed by visual inspection. The objective of this study was to develop and validate a new computer algorithm for quantifying regional left ventricular wall thickening on ECG-gated SPECT images. METHODS: Regional wall thickening was measured from count density changes during the cardiac cycle observed in 24-sector circumferential count distribution profiles generated from each of 8 frames of an ECG-gated SPECT study. Wall thickening was expressed as the percent count increase during systole relative to end diastole. The program was tested in a phantom simulation and in patient studies consisting of a pilot study (n = 40) and a validation study (n = 33). In the phantom study varying degrees of wall thickening were simulated. The pilot study included 20 normal subjects with low likelihood (<3%) of coronary disease and 20 patients with prior myocardial infarction. Mean wall thickening - 2 standard deviations, measured in normal subjects, defined the lower limit of normal wall thickening. This criterion was tested in the validation study in 13 normal subjects and 20 patients with prior myocardial infarction. Abnormal wall thickening was characterized by extent (percent of circumferential profile) and severity (minimal thickening). RESULTS: The phantom study showed excellent linear correlation between wall thickening computed by the new software and actual wall thickening (r = 0.98). Interobserver and intraobserver reproducibility of quantitative assessment of minimal wall thickening were excellent (r = 0.98 and 0.99, P < .001). Regional wall thickening varied considerably from apex to base in the same ventricle among normal subjects. The average lower limit of normal wall thickening was 25% to 30% at the apex, 19% to 24% in the mid-ventricle, and 13% to 20% at the base of the left ventricle. In the validation study 11 of 13 normal subjects had wall thickening profiles within the pre-defined normal range. All 20 patients with prior myocardial infarction had abnormal regional wall thickening. Minimal regional wall thickening in the infarct areas was 5.4% +/- 5.5%, compared with 30.1% +/- 9.1% wall thickening in comparable anatomic areas in normal subjects (P < .001). CONCLUSION: Regional wall thickening can be quantified reliably from regional count density changes during the cardiac cycle on ECG-gated SPECT images. The new software measured the extent and severity of abnormal regional wall thickening relative to normal files. The method is highly reproducible. Clinical validation showed good differentiation between normal subjects and patients with prior infarction. Quantification of regional wall thickening may enhance diagnostic accuracy and reproducibility of interpretation of gated SPECT imaging.