Radionuclide cardiac volumes: effects of region of interest selection and correction for Compton scatter using a buildup factor.

The effects of region of interest (ROI) selection and correction for Compton-scattered photons using a buildup factor on radionuclide left ventricular volumes calculated by the Links method were compared in 19 humans with contrast ventriculography and in phantoms. Three different methods of ROI selection were compared: a manual ROI, a second derivative ROI and a 50% count-threshold ROI. In phantoms without Compton scatter correction, volumes were overestimated by 30% (manual ROI), 20% (derivative ROI) and 1% (count threshold ROI). In subjects, results without Compton scatter correction were similar with overestimates of 50% (manual ROI) and 20% (derivative ROI) and an underestimate by 3% (count threshold method). Correction for Compton-scattered photons with the use of a phantom-derived buildup factor resulted in improved accuracy for the manual ROI (+15%) and the derivative ROI (0%). A 50% count threshold ROI following interpolative background subtraction allows the accurate calculation of cardiac volumes without the need for scatter correction, while a second derivative ROI method requires a correction for Compton scatter with the use of a buildup factor.