Effect of Variation in Heart Size on Quantitative Thallium SPECT

Quantitative single-photon emission tomographic (SPECT) thallium 201 (201Tl) cardiac studies are frequently used to improve test accuracy. One common technique uses polar mapping of relative tracer distribution to compare patient data to emplates from "normal" patient data. Commercial software does not correct for cardiac size variations. This study's purpose was to determine the effect of uncorrected variations in heart size on the variance of the normal population, which is the basis of template significance levels. Twenty-one male volunteers with low probability (< 5%) for coronary disease underwent SPECT-Tl stress test. Data analysis with and without size correction was performed. Data were corrected for size by use of a three-dimensional compression/expansion algorithm. Quantitative data were generated by radial search to maximum activity in short-axis reconstructions. The mean activity and its variance was calculated from 288 points encompassing the myocardium. Mean and variance differences were analyzed by paired t test. There was no difference in activity means (pT > .9), indicating that size correction does not distort the data. The corrected data had, however, significantly lower variance (pT < .0001). This shows that ventricular size variation contributes significantly to "normal" template variance and may adversely affect diagnostic accuracy in the absence of volume correction.