Phantom validation of a new method for quantifying end-diastolic volume (EDV) and ejection fraction (EF) from SPECT equilibrium radionuclide angiocardiography (ERNA)

Objectives The Massardo method for quantifying EDV and EF from 2D planar ERNA assumes a spherical shape of the left ventricle (LV), which is not a realistic estimate. We developed & validated a new (Yale) method in phantom using 3D SPECT w/ & w/o CT attenuation correction (AC).

Methods Six double-walled cylindrical phantoms w/ inner volumes of 95, 74, 55, 39, 28 &19 mL were used. The inner & outer spaces of each cylinder were respectively filled w/ 2.5μCi/mL & 0.25μCi/mL of Tc99m & submerged in a larger cylinder w/ water. Images were acquired for 6 min & reconstructed via MLEM w/ AC & w/o AC (NC), w/ & w/o Green regularization. No scatter correction & post-filtering were applied. LV slices were auto-defined & six sets of SPECT-reprojected images were created/aligned/combined to generate an 11-bin pseudo ERNA. EDV images were quantified by Yale & Massardo methods. EF was calculated from background-corrected (EDcounts-EScounts)/EDcounts. Quantified EDV & EF and their estimation errors (EE) were compared to the true phantom values (EDV= 95 mL; EF= 80%).

Results AC- & NC-quantified EDVs via both methods were significantly different (Table). EE of EDV w/ NC & AC by Yale method were small. Massardo method markedly underestimated EDV. Difference between AC & NC quantified EF was small, but statistically significant (p= 0.05; n= 10) with similar EE (AC:-10.6 ± 1.58 % vs.NC: -9.3 ± 1.49 %; p= ns; n= 10). Variation coefficients (VC) of NC & AC quantified EFs were small & identical (VC= 2%). Quantified EF underestimated true EF by ~10%, presumably due to the suboptimal background correction.

Conclusions EDV quantification by Yale's method was more accurate than that by Massardo's. AC had minimal effect on EDV quantification in both methods. The precision of EF quantification was excellent, though the accuracy was fair.