Quantification of left ventricular volumes and ejection fraction from gated 99mTc-MIBI SPECT: MRI validation of the EXINI heart software package

Abstract

The aim of the study was to validate the accuracy of the EXINI heart software (EXINI) package in assessing left ventricular end-diastolic/systolic volumes (EDV, ESV) and ejection fraction (LVEF) from gated (99m)Tc-MIBI single-photon emission tomography (SPECT). Cardiac magnetic resonance imaging (cMRI) was used as reference. Furthermore, effects of perfusion defects and image quality in SPECT on correlation between gated SPECT and magnetic resonance imaging were investigated.

Methods: Seventy patients were examined using gated SPECT (rest study, eight gates per cardiac cycle). EDV, ESV and LVEF were calculated from gated SPECT using EXINI. Directly before or after SPECT, cMRI (20 gates cardiac per cycle) was performed. EDV, ESV and LVEF were calculated using Simpson's rule. Perfusion defects were quantified using the summed-rest-score (SRS). Total number of myocardial counts were used to rate image quality.

Results: Correlation between results of gated SPECT and cMRI was high for EDV (R = 0.89) and ESV (R = 0.94) and good for LVEF (R = 0.78). ESV (EXINI 54 +/- 31 ml versus cMRI 57 +/- 34 ml) and LVEF (EXINI 62.9 +/- 11.7% versus cMRI 60.6 +/- 13.9%) did not differ significantly whereas EXINI overestimated EDV significantly compared with cMRI (EXINI 144 +/- 41 ml versus cMRI 137 +/- 36 ml; P<0.005). No correlation was found between absolute differences of the results given by gated SPECT and cMRI and SRS or total myocardial counts (R < 0.18).

Conclusion: End-diastolic volume, ESV and LVEF calculated from gated SPECT using EXINI agree with cMRI over a wide range of values. Correlation between both the methods was good for EDV and ESV, and acceptable for LVEF. No relevant influence of image quality or SRS on the accuracy of EXINI results was found.