Multidetector computed tomography evaluation of left ventricular volumes: sources of error and guidelines for their...

Although multidetector computed tomography (MDCT) is known to overestimate left ventricular (LV) end-systolic and end-diastolic volumes (ESV, EDV) compared to magnetic resonance imaging reference, the potential sources of error have not been thoroughly investigated.

We sought to quantitatively assess the effects of the number of reconstructed phases and number of slices used for volume calculation on the accuracy of LV volume measurements.

MDCT images obtained in 28 patients (Philips Brilliance 64) were reconstructed at 10, 20, 33, and 100 phases per cardiac cycle. For each number of phases, ESV was measured between aortic valve closure and mitral valve opening and normalized by reference ESV measured at 100 phases/R-R. Both reference ESV and EDV were measured using 20 and separately 10 fixed-thickness slices. Reproducibility was assessed using repeated measurements.

In 16 of 28 patients, the timing of end-systole varied with increasing number of reconstructed phases, resulting in a gradual decrease in ESV from 118 +/- 20% of reference ESV to 100 +/- 0%. Reduction in number of slices caused a significant increase in EDV and ESV (4.2 +/- 3.2% and 6.4 +/- 5.5%, respectively), roughly twice the corresponding intraobserver variability (2.5 +/- 1.5% and 3.8 +/- 2.4%).

Misidentification of end-systole due to insufficient number of reconstructed phases significantly affects ESV measurements. Also, the number of slices used for volume calculation affects both ESV and EDV beyond intermeasurement variability. To ensure accurate quantification of LV volumes, reconstruction at time intervals smaller than 5% of the RR-interval (>20 phases/cardiac cycles) and tracing endocardial borders in >10 slices are recommended.