Measurement of left ventricular volume after anterior myocardial infarction: comparison of magnetic resonance imaging, echocardiography, and radionuclide ventriculography

We have compared echocardiography (echo) and radionuclide ventriculography (RN-V) with magnetic resonance imaging (MRI) for the measurement of left ventricular (LV) volume and ejection fraction. Seventy asymptomatic patients were studied up to 12 days after first Q wave anterior myocardial infarction and again after 6 months. Each patient had LV volume measured by all three techniques within 24 hours of each other on each occasion. LV end-systolic and end-diastolic volume index (LVESVI and LVEDVI) and LV ejection fraction (LVEF) were measured using the modified Simpson formula (echo), a counts-based method (RNV), and a multislice area summation method (MRI). Radionuclide volumes were measured both with and without correction for attenuation of isotope. Echocardiography overestimated LV volume compared with MRI. Mean (SD) differences (echo-MRI) were: LVEDVI + 10.6 ml/m(2) (16.8), LVESVI +13.7 ml/m(2) (12.9), LVEF -8.5% (11.2). RNV underestimated both volume and ejection fraction compared with MRI. Mean differences (RNV-MRI) were: LVEDVI -25.4 ml/m(2) (23.8), LVESVI -5.0 ml/m(2) (18.6), LVEF -13.8% (10.4). Variability in the difference between echo and MRI and between RN-V and MRI was very similar for LVEF (coefficient of variation 23.9% echo, 22.2% RNV) but there was greater variability in the radionuclide than the echo measurements of absolute volume. Variability of the radionuclide measurements was reduced by not correcting for attenuation, and this finding may improve the radionuclide technique for serial measurements of percentage change in volume. Long-term inter-study reproducibility of MRI for LVEF (coefficient of reproducibility) was 10.9%, for echo it was 10.6%, and for RNV it was 14.6%. We conclude that measurements of LV volume depend on the method used and are not interchangeable. Echocardiography agrees more closely with MRI than RNV for the measurement of absolute volume, but the two techniques are similar for the measurement of LVEF.