Dependence of ejection fraction results on choice of algorithms for a cardiac phantom.

Ejection fraction results obtained from the Vanderbilt rotating cardiac phantom exhibit a strong dependence on the algorithms used to derive the time--activity curves and calculate the ejection fraction from those curves. This problem is caused because the background chamber is located behind the rotating attenuator which is used to produce the pseudo ejection fraction. As a result, the counts due to background in both the regions on the 'left ventricle' and the area assigned for background correction do not remain constant during the 'cardiac cycle'. This leads to a breakdown in the ejection fraction algorithms. Given the same data, but changing the methods by which the time--activity curves and ejection fractions are derived, ejection fraction results varying from a minimum of 45% to a maximum of 69% have been demonstrated for the 50% attenuator. Although the Vanderbilt cardiac phantom is simple to operate it must be used with caution if it is used with the intention of calibrating a particular system or comparing different protocols.