Comparison of four commercial software for the analysis of cardiac volume and ejection fraction when applied to PET ammonia images

1714 Objectives In this work we analyzed data from ammonia PET studies using four different commercial software validated for gated SPECT images: Myometrix (MYO), QGS, Emory Cardiac Toolbox (ECT), CardIQ Physio (CIQP). Methods 28 pts with different pathologies underwent rest and vasodilatatory stress study by using 13NH3 and 3D-PET. The acquisition protocol was: injection of 370 MBq of 13NH3 and a 3D-PET list mode scan (20 min). PET data (last 15 min) were processed into 12 cardiac gates and reconstructed with a 3D-OSEM algorithm also accounting for TOF and PSF. PET images were analyzed with the four software. Paired t-test was used to assess differences in the results. Rest and Stress were analyzed separately as well as altogether to increase statistical power. Comparisons are reported for end-diastolic volumes(EDV) and ejection fractions(EF). To assess inter-program variabilities we computed coefficients of variations(CV) and standard deviations (SD) of paired comparison. Results Mean EDV[Rest/Stress] for MYO were:129/132ml,ECT: 112/115ml,QGS:111/116ml, CIQP 137/141ml. Mean EF were[Rest/Stress]:58/60%,64/65%,51/56% and 55/55%. The paired comparisons between software are reported as: MYO vs ECT, MYO vs QGS, QGS vs ECT, MYO vs CIQP, ECT vs CIQP and QGS vs CIQP. The mean differences in EDV were: 14%±1%,14.5%±1%,0.3%±1%,- 8%±1%,- 22%±1%,- 22%±1%.The CV were:6.8%,6.0%,6.6%,5.8%, 7.7%,7.0%.The EF mean differences(absolute differences of the EF)were:-5.7%±1%,,5.5%±1%,11%±1%,4.4%±0.5%,10%±1%,-1%±1%. The CV between the programs were: 6.0%,6.0%,6.0%,4.5%,6.5%,6.7%. Conclusions The inter-software variabilities has been assessed and were found to be limited to about 7% for EDV and 6% for the EF. Given the high mean differences seen between the software care should be taken when comparing results to normalcy limits.