Repeatability of estimates of left-ventricular volume from blood-pool counts: concise communication.

Radionuclide ventriculography permits nongeometric calculation of ventricular volume. Accurate and reproducible determination of left-ventricular (LV) blood-pool counts is necessary to perform this calculation. Furthermore, to make serial volume determinations one must know the half-time of in vivo blood-pool activity. We compared five methods of LV count determination in nine patients. Interpatient and intrapatient variability of the in vivo half-time of Tc-99m-labeled red blood cells (RBCs) was measured. Left-ventricular count determinations, derived from temporally and spatially smoothed images using a second-derivative algorithm to identify the LV region of interest (ROI), are less variable than those based on manual ROI determinations. The mean in vivo half-time of Tc-99m RBCs is 4.1 hr, and there is significant interpatient (0.9 +/- 0.8 hr) and intrapatient (1.0 +/- 0.9 hr) variability. These findings should be considered in the determination of serial, relative ventricular volume by radionuclide ventriculography.