Verification of a varying threshold edge detection SPECT technique for spleen volume: a comparison with computed tomography volumes.

SPECT enables quantitation of organ volume with radionuclide techniques using threshold edge detection methods. Previous phantom studies showed that a negative correlation exists between volume and threshold value. In those studies, the use of calibration curves were believed to correct for volume dependence on threshold values. The aim of this study was to evaluate the accuracy of spleen volume determination in 20 patients with SPECT by employing a varying threshold edge detection technique with volumes derived from CT. All patients had both radionuclide and CT examinations that were reconstructed with a filtered backprojection algorithm. During SPECT reconstruction, transverse slices were obtained with attenuation correction (Method A) and without attenuation correction (Method B). CT volumes were calculated from manually drawn regions of interest, whereas SPECT volumes were calculated with an automated algorithm using previously determined calibration curves. A confidence interval for calculated SPECT volumes also was calculated because of possible errors in the threshold value. The spleen volumes studied ranged from 91.2 ml to 1660.1 ml. Regression analysis yielded equations of CT = 0.97 SPECT + 7.07 (r = 0.996) and CT = 1.05 SPECT - 19.25 (r = 0.990) between CT and SPECT spleen volumes with a standard error of the y estimates of 31.10 ml and 54.47 ml, respectively. A mean percentage difference of 10.5% +/- 7.6% and 11.4% +/- 6.6% in spleen volume was obtained for Methods A and B in comparison with CT spleen volumes. The threshold value varied between 40.9% and 32.4% for Method A and between 41.2% and 28.5% for Method B because the spleen volume is increased. The varying threshold edge detection technique described in this paper can be implemented successfully in the clinical setting.