LVEF is overestimated by conventional SPECT in patients with small hearts, and the incidence of small heart size is high (nearly 80% in Asian women) [14]. On the other hand, IQ-SPECT has advantages such as fast acquisition time, high resolution, and low injection volume. Because conventional SPECT is still more common, we compared the two methods in patients with small hearts to determine which method was more similar to CMR. The objectives of this study were to evaluate EDV, ESV, and LVEF obtained using conventional SPECT and IQ-SPECT, with CMR as the gold standard, and to compare the LVEF detected using IQ-SPECT with different reconstruction parameters between those with small hearts and those with normal-sized hearts. The results suggested that IQ-SPECT might show some advantages for patients with small heart size.

CMR was selected as the reference method because it is the accepted standard for measuring global function [20]. Furthermore, volumetric accuracy was ensured by using high tissue contrast for the definition of the endocardial border [21, 22]. There are more contraindications to the use of CMR than SPECT, and there is value in comparing the LVEF detected by CMR and SPECT in patients with small hearts.

When the IQ-SPECT system was introduced in 2011, the manufacturer recommended using the Siemens Flash3D iterative reconstruction algorithm with 15 I and 2 S for the processing of gated images. In our work, we observed that the mean LVEF determined by IQ-SPECT was smaller than that determined by conventional SPECT. Hence, we compared EDV, ESV, and LVEF obtained by IQ-SPECT, conventional SPECT, and CMR. The closest agreement was observed between the LVEF detected by IQ-SPECT and that examined by CMR.

We compared 6 groups of reconstruction parameters in patients with small hearts and normal hearts to determine the effect of altering these parameters. The best Flash3D iterative reconstruction algorithm, with CMR as the standard, differed between small-heart and normal cardiac chambers. The optimal reconstruction parameters for patients with a small heart were 2 S and 15 I, and the optimal reconstruction parameters for those with normal hearts were 3 S and 10 I. In our study, there were no differences in the LVEF detected using IQ-SPECT and CMR. It is important to choose different reconstruction parameters for heart chambers of different sizes.

Nakajima et al. [14] found that small heart size (EDV < 20 ml) had a prevalence of 74% in Japanese women and 13% in Japanese men. Kakhki et al. [17] also found that 85.4% of Iranian subjects had an ESV of < 25 ml (94.9% of women and 11% of men). ESV is underestimated in those with small LV volume, and LVEF is overestimated, with a greater error in females [11, 23,24,25,26,27]. Indeed, due to the limited spatial resolution of gamma cameras, the opposite endocardial edges of the LV overlap; thus, the ventricular cavity may become almost completely indistinct, especially at end-systole. QGS examination was shown to overestimate the ejection fraction in patients with small hearts, especially when the EDV was < 70 ml or the ESV was < 25 ml [12, 17, 23, 28, 29]. This finding implied that different thresholds must be used for subjects with normal and abnormal heart sizes.

The ability of gated conventional SPECT to measure LVEF, segmental wall motion, and absolute LV volumes has been extensively validated in head-to-head comparisons with clinically proven imaging methods such as planar radionuclide ventriculography, contrast ventriculography, CMR, and conventional SPECT imaging [12, 17, 23, 28, 29]. IQ-SPECT allows a significant reduction in the administered dose and acquisition time for myocardial perfusion imaging. Many studies have investigated the consistency of LVEF detected by conventional SPECT and IQ-SPECT, but the results have been inconsistent [3,4,5,6, 8, 30].

Pirich et al. [30] reported a significant difference in functional parameters derived from IQ-SPECT and conventional SPECT both after stress and during rest. The mean LVEF obtained by IQ-SPECT was reduced by 8%. The average LVEF after stress as measured by IQ-SPECT was 49.2 ± 13.0%, whereas the value obtained using conventional SPECT was 57.1 ± 12.5%. The average LVEF values as measured by IQ-SPECT and conventional SPECT during rest were 47.2 ± 12.8% and 56.4 ± 14.5%, respectively. Havel et al. [3] reported that the average LVEFs determined by IQ-SPECT and conventional SPECT were 54.1 ± 14.0% (1 S; 30 I; 14-mm full width at half maximum (FWHM) Gaussian filter) and 61.9 ± 12.2%, respectively, which were similar to our results (including all patients). Nevertheless, Yoneyama et al. [6] reported that the mean LVEF values obtained with QGS from IQ-SPECT were higher than those obtained from conventional SPECT for individuals with small and normal hearts. The average LVEF values of all patients as measured by IQ-SPECT and conventional SPECT were 68.4 ± 15.2% and 65.4 ± 13.8%, respectively. The average LVEF of patients with a small heart as measured by IQ-SPECT and conventional SPECT were 79.5 ± 8.3% and 75.0 ± 9.6%, respectively. Matsutomo et al. [8] reported that the EDV, ESV, and LVEF obtained by IQ-SPECT (1 S; 30 I; 13-mm FWHM Gaussian filter) did not significantly differ from those obtained by conventional SPECT. Although the mean IQ-SPECT-measured LVEF in that study was higher than the LVEF measured by conventional SPECT, the difference was not significant (68.3 ± 12.1% vs. 64.8 ± 11.8%, P = 0.269). Twenty-five patients were included in the study, with only seven being female, and the effect of small heart size on the IQ-SPECT system was not clarified. In the present study, we included individuals with small hearts as well as individuals with normal hearts. The results from the two groups were consistent. The mean LVEF detected by IQ-SPECT was lower than that detected by conventional SPECT for both groups.

Yoneyama et al. [6] reported EDV, ESV, and LVEF obtained using conventional SPECT, IQ-SPECT, and echocardiography. They demonstrated a good to excellent correlation between these methods. The present study found that EDV, ESV, and LVEF obtained from conventional SPECT and from IQ-SPECT showed a good to excellent correlation. In addition, we found that the LVEF measured by IQ-SPECT agreed more closely with the result of CMR detection than the LVEF measured by conventional SPECT for small hearts (62.1 ± 7.8% vs. 64.6 ± 8.8%, P = 0.120) and normal hearts (50.6 ± 4.3% vs. 53.2 ± 5.8%, P = 0.056).

The parameters of cardiac function measured using myocardial perfusion imaging are significantly influenced by the reconstruction algorithm. Nakajima et al. [14] observed that small heart size had a prevalence of 74% in Japanese women and 13% in Japanese men. In patients with small hearts, the true volume is underestimated, and this effect is greater for ESV than EDV, leading to an increase in the apparent LVEF. This small-heart effect, seen in several studies, is caused by the optimization of the SPECT reconstruction method for a myocardial wall with poor resolution [9,10,11,12,13, 23, 31, 32]. Nakajima et al. [23] reported that the use of a high cut-off frequency for the SPECT filter, high system resolution and proper zooming could improve gated SPECT quantification for small hearts. An improvement in spatial resolution could significantly decrease the small-heart effect.

There is currently no unified standard for optimal OSCGM parameters. Kenda et al. [33] reported that the optimal reconstruction minimization for IQ-SPECT was 1 S and 30 I. Based on these parameters, the measured LV volume was similar to the actual volume. They performed the evaluation using an RH-2 cardiac phantom (Kyoto Kagaku Co., Ltd.) containing a solution of 99mTc. Ceriani et al. [34] reported that the QGS program was able to calculate the LVEF correctly when used in conjunction with an optimized 3D OSCGM algorithm (8 S, 10 I, and an FWHM of 10 mm) but that it resulted in an underestimation of LV volumes. Duarte et al. [35] reported a more precise estimation of the quantitative parameters with OSCGM, especially with the combination of 2 I × 10 S and 2 I × 12 S. Nevertheless, they were less accurate in a validation study using a beating-heart volume phantom (ESV of 33.5 mL and EDV of 108.5 mL) than in studies of patients with small hearts. The present study found that the optimal minimization of reconstruction for IQ-SPECT was 5 S and 12 I with an FWHM of 8 mm. With these settings, SPECT LVEF was similar to that of CMR LVEF. The ESV was similar to that detected by CMR with 8 S and 18 I with an FWHM of 8 mm. The present study also showed that the optimal reconstruction minimization for IQ-SPECT was 3 S and 10 I with an FWHM of 8 mm. Based on these parameters, the LVEF for a normal heart was similar to that detected by CMR.

The sample size was small. Patients with large hearts were not evaluated to analyse reconstruction parameters. Another limitation was that this was a single-centre study.