Variations in cardiac cycle length during ECG-gated radionuclide ventriculography cause errors in measurement of left ventricular (LV) diastolic function. We used a computer model to investigate the relative accuracy of forward gating (FG), backward gating (BG), and three methods of combined forward-backward gating (FBG). LV time-activity curves (TAC) were simulated with total cycle lengths randomly varied (1 s.d. = +/- 10%). To match the known differential effects of cycle-length variation on

more »

... arious parts of the cardiac cycle, 90% this variation was placed in diastasis and 10% in the rapid filling phase. Three hundred cycles of TACs were summed to form one "study". A total of 25 studies were performed and systolic and diastolic parameters were determined for each. These were compared to those of the mean length TAC. With a 20% cycle-length acceptance window, error in peak filling rate by FG was 0.7% root-mean-squared (RMS) compared to 18.5% RMS by BG. With a 5% window, RMS error by FG was 0.2% and 2.4% RMS by BG. Combined FBG algorithms gave intermediate results. We conclude that the simplest and most accurate method of analysis is to acquire forward gated TACs with appropriate cycle-length windows.