Comparison of Left Ventricular Phase Parameters Analysis between Two Software Programs in Patients with Normal Gated Single-photon Emission Computed Tomography-Myocardial Perfusion Imaging.

BACKGROUND: Phase analysis can be easily performed by different software to assess the left ventricular dyssynchrony from gated single-photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI) with high precision. However, the normal values of histogram bandwidth (HBW) and phase standard deviation (PSD) and their comparison using different programs have not been fully elucidated and actively being evaluated at present. The aim of this study was to determine the phase analysis parameters values and to compare the phase indices of two commonly used programs in a group of patients with normal gated SPECT-MPI.
METHODS: Phase parameters were retrospectively evaluated in 138 consecutive nondiabetic patients having a normal gated SPECT-MPI using the quantitative gated SPECT (QGS) and Emory Cardiac Toolbox (ECTb) software. HBW, PSD, and phase entropy were calculated separately using both programs.
RESULTS: The fair correlation between software programs was observed. HBW and PSD in QGS and ECTb were 26.20 ± 9.7 and 25.46 ± 8.0 (r-value. 56, SEE 6.65) and 6.64 ± 2.5 and 7.65 ± 2.5 (r = 0.54, SEE 2.14), respectively. The value of phase entropy in QGS program was 45.08 ± 6.3. A fair correlation between phase entropy and PSD in QGS was observed (r = 0.44, 95% confidence interval-0.29-0.56).
CONCLUSION: Phase analysis parameters derived from gated SPECT-MPI in patients with normal myocardial perfusion are program dependent and may differ. The results cannot be interchangeably used in the same patients.

Introduction

Left ventricular dyssynchrony (LVD) can be used as a prognostic and risk stratification marker in patients with ischemic, nonischemic cardiomyopathy and end-stage renal disease.[123] Phase analysis using gated single-photon emission computed tomography myocardial perfusion imaging (SPECT-MPI) as a diagnostic tool for the evaluation of LVD was introduced in 2005 by Chen et al.[4] LVD can be assessed by multiple imaging modalities such as echocardiography, magnetic resonance imaging, and equilibrium radionuclide angiography. In addition to these techniques, phase analysis by gated SPECT-MPI provides excellent reproducibility and repeatability though data on phase analysis derived from gated SPECT MPI is still sparse.[5] In addition, SPECT-MPI allows simultaneous assessment of left ventricular perfusion, function, and localization of myocardial scar (if present any) providing a complete package of information. In the evaluation of LVD, phase analysis of gated SPECT-MPI can be performed by commercially available different types of software. The two most commonly and widely used software in gated MPI are Emory Cardiac Toolbox-SyncTool (ECTb-SyncTool; Emory University/Syntermed, Atlanta, Georgia, USA) and quantitative gated SPECT (QGS; Cedars Sinai Medical Centre, USA). However, the comparison between them has not been fully evaluated yet. The aim of present study was to determine the normal phase analysis parameters and to compare them using these two different software programs in a group of patients with normal myocardial function.

Methods

Study population

This retrospective study was conducted at the Nuclear Cardiology Department of Advance Cardiac Centre, PGIMER, Chandigarh. Patients with normal gated stress SPECT-MPI on a 17-segments, 5-point (0–4) model (i.e., (SSS) summed stress score ≤3 and left ventricular ejection fraction [LVEF] >60%) with no regional wall motion abnormalities were included in the study. Patients did not have any underlying cardiac diseases at the time of imaging. Patients with the previous history of cardiac disease, diabetes mellitus, coronary intervention, inappropriate arrhythmia for gating, wall motion abnormality, and the presence of any perfusion defect on gated stress SPECT-MPI were excluded from the study. A total of 138 consecutive patients (71 female and 67 males) following inclusion criteria were included, and data were analyzed separately by the two-experienced nuclear medicine physicians who were blinded to the study.

Gated myocardial perfusion single-photon emission computed tomography acquisition

All patients underwent Tc-99 m 1-day stress- first/stress-only gated SPECT-MPI protocol according to the American Society of Nuclear Cardiology guidelines being routinely followed in our department. The stress dose (6.3 MBq/Kg) of 99mTc sestamibi (hexakis-6-methoxyisobutylisonitrile) was given on the basis of weight of patient in the SPECT-MPI study.[6] The rest study was not performed in these patients as stress perfusion imaging was normal (SSS ≤3 and LVEF >60%). Sixty-six patients underwent exercise stress test according to Bruce protocol out of 138 patients and remaining 72 patients underwent pharmacological stress test (adenosine n = 66 and dobutamine n = 6) depending on the patient's clinical condition.

Gated SPECT-MPI acquisition was performed on a dual-head camera system (Philips Bright view XCT, Philips Medical Systems, Milpitas, CA, USA) 30–45 min after the stress procedure in the supine position. Stress gated images were acquired using a 15% window centered over the 140 KeV photopeak of 99mTc with parallel hole, low energy, and high-resolution collimator. electrocardiogram-gated SPECT imaging was performed with eight frames per cardiac cycle, using a 100% beat acceptance window. A total of 64 projection (32 steps/per detector, 3° steps) of 25 sec/projection were acquired over 180° from 45° right anterior oblique position to 135° left posterior oblique position.

Image processing

SPECT nongated projection images were reviewed in cine mode in all cases to assess the patient's movement, sources of potential attenuation artifacts, and gastric activity. The raw images (both gated and nongated datasets) were reconstructed using manufacturer provided Astonish (iterative reconstruction and resolution recovery) software (AutoSPECTPlus™, Philips Medical Systems) by an experienced technologist, who was blinded from this research project.

Assessment of phase analysis

Phase analysis was performed with ECTb-SyncTool version 3.2 and QGS version 7.2 software separately by two experienced nuclear physicians in blinded manner. The analyses were performed using standard software settings of ECTb and QGS. ECTb used anatomically based on three-dimensional model for ventricular edge detection, assuming 1-cm LV myocardium thickness at the end-diastole, whereas ellipsoid shape and iterative process to fit the myocardial wall was used in QGS software. Fourier harmonic functions were used to approximate the regional wall thickness changes over the cardiac cycle and to calculate the regional onset-of-mechanical contraction phase. After the onset-of-mechanical contraction phases were obtained three-dimensionally over the left ventricle, a phase distribution map was formulated to represent the degree of the LVD. The following phase analysis parameters were recorded: Peak phase, phase standard deviation (PSD, unit: Degree), and 95% width of histogram bandwidth (HBW, unit: Degree). The 95% width includes practically full HBW with the exclusion of likely outlier values. In addition, QGS software provides another index of dyssynchrony, phase entropy, which is normalized to its maximum value and reported as a percentage ranging from 0% to 100% from complete order to disorder. For outcome analysis, only PSD and HBW have been used as quantitative indices of LV dyssynchrony. PSD, HBW, and entropy increase as the LV mechanical synchrony worsens.

Statistical analysis

Statistical analysis was performed using SPSS software (version 22.0, NY, USA: IBM Corp.) Continuous variables were expressed as mean ± SD, compared using the paired and unpaired Student's t-test or Wilcoxon rank test as appropriate. Categorical variables were expressed as number and percentage. Correlations between continuous variables were assessed by the Spearman's Rank correlation test and Pearson coefficient. Statistical significance was set at a level of <0.05.

Results

Patient's characteristics

Baseline characteristics of study patients are described in Table 1.

Table 1

Patient characteristics

One hundred thirty-eight patients (mean age 54.9 ± 11.8 years, 71 females) were included in the study. Ninety-one patients (66%) had hypertension and 22 (16%) were alcoholic while elevated lipid profile was observed in 21 (15%) patients. The patients with hypertension and dyslipidemia were well under control with medication and modified lifestyle. The dataset consisted of patients with satisfactory image quality and gating.

Phase analysis results

The phase analysis value of both the software is given in Table 2. The mean values ± SD of poststress HBW were 26.20 ± 9.7 and 25.46 ± 8.0 in QGS and ECTb, respectively. The mean values of PSD were 6.64 ± 2.5 and 7.65 ± 2.5 in QGS and ECTb, respectively. A fair correlation for both PSD and HBW (r = 0.54; SEE 2.14 and. 56; SEE 6.65, respectively) was observed among two software programs [Table 2]. A representative dataset of patient with normal SPECT-MPI showed different PSD and HBW obtained with different programs [Figure 1]. The value of phase entropy, provided only by QGS software was 45.08 ± 6.3. On further analysis, we found a fair correlation between phase entropy and PSD in QGS software (r = 0.44, SEE 5.74, 95% confidence interval-0.29–0.56) [Figure 2]. Gender-based analysis of various phase parameters was also done and males showed larger phase deviations (PSD, HBW, and phase entropy) [Table 3] as compared to female counterpart,; however, the difference was statistically not significant.

Table 2

Phase analysis result and correlation between two different software (Emory Cardiac Toolbox. SyncTool and quantitative gated single photon emission computed tomography)

Figure 1

Relationship of phase standard deviation (a) and histogram bandwidth (b) using Emory Cardiac Toolbox-SyncTool and quantitative gated single-photon emission computed tomography software. Correlation between phase standard deviation and phase entropy (%) using quantitative gated single-photon emission computed tomography software (c). The black line represents the central fit line and dotted red line indicates 95% confidence limits for the regression line

Figure 2

Represents the dataset of patient number 114. Splash images (a) does not reveal any perfusion defect in both Emory Cardiac Toolbox and quantitative gated single photon emission computed tomography software. (SSS) summed stress score score and left ventricular ejection fraction was <4 and >60%, respectively, for this patient. Phase analysis done with Emory Cardiac Toolbox (b) revealed phase standard deviation of 9.8 and histogram bandwidth was 36. However, when same dataset was analyzed in quantitative gated single-photon emission computed tomography software (c) parameter observed were 12 and 42 respectively, revealing a wide variation in phase parameters between two different software

Table 3

Phase parameters in male and female patients

Discussion

Gated SPECT-MPI is now widely used for the assessment of LVD and as a response indicator to cardiac resynchronization therapy (CRT).[8] Chen et al. in 2005 were first to propose the normal database value using ECTb software for the assessment of LVD.[4] Subsequently, many studies had published normal values of stress PSD and HBW in control groups except rest phase parameters in the study done by Nakajima et al. [Table 4] using ECTb software.[59101112] Interestingly, all the values obtained from different control groups were different. Similarly, normal value of PSD (7.65 ± 2.5) and HBW (25.46 ± 8.0) obtained in the present study using ECTB software were also different from previously established values in the literature. The prevalence of cardiovascular risk factors such as diabetes, hypertension, and dyslipidemia in control groups could potentially affect the phase analysis values.[13] Furthermore, the selection criteria of control group and methodology (acquisition/processing protocols, software algorithm, and radionuclides) in these studies were different. Trimble et al.[9] in their study included patients with atrial fibrillation in the control group. Whereas, the normal database first proposed by Chen et al.[4] derived from a standard Tl-201/ 99mTc sestamibi dual isotope rest/exercise protocol. In the present study, we included patients in whom gated stress SPECT-MPI was normal with adequate LVEF >60% and Tc-99 m 1-day stress- first/stress-only gated SPECT-MPI protocol was used. These differences in the selection criteria of normal group and methodology could possibly explain different normal values of PSD and PHB obtained from different studies.

Table 4

Established normal database of phase analysis (phase standard deviation and histogram bandwidth) using Emory Cardiac Toolbox software

Chen et al.[4] in their study proposed higher values of phase analysis parameters in male s compared to females, which was later supported by the data published from the Indian[11] and Japanese population.[12] The explanations provided for this includes the presence of greater body mass index and LV mass in male in comparison to females and a concept based on electrophysiological studies showing that the large myocardial mass in male takes longer time to depolarize.[14] Since phase analysis is a count-based technique, it could be influenced by count density, so if similar dose is injected, larger heart will accumulate less myocardial count per unit volume. Lower counts per pixel will result in the higher noise and potential measurement errors leading to higher PSD and HBW indices.[15] Only standard radiotracer doses irrespective of patients' weight were used in the stress gated MPI in both the studies.[1112] However, in the present study, the dose used was based on patient's weight which could have eliminated the effect of count density, though we found the trend for higher values for both PSD and HBW in males than females. This may be partly explained by the known impact of processing software on measurement of LV volume and LVEF.

Phase analysis parameters in the normal population available in the reported literature are derived from the ECTb mostly, and phase analysis parameters for normal population using QGS software had not been well established. However, Boogers et al.[16] reported the cutoff value of 72.5 and 19.6 for HBW and PSD, respectively, with sensitivity and specificity of 83% and 81% in patients responding to CRT using QGS software. Similarly, Rastgou et al.[17] in 2014 calculated the mean values of HBW (112.2) and PSD (28.7) using QGS software in patients with heart failure. Recently Nakajima et al.[12] had evaluated phase analysis parameter using four different software in patients with normal SPECT-MPI. The parameter obtained from QGS software for HBW, PSD, and entropy are 25.0 ± 8.9, 6.2 ± 3.0 and 27.8 ± 7.8, respectively. Like ECTb results, parameter in QGS software also varies significantly depending on the patient group. The normal value of phase analysis parameter HBW and PSD obtained in QGS software in our study were 26.20 ± 9.7 and 6.64 ± 2.5, respectively, which did not differ significantly from the published literature. Recently Okuda et al.[18] in 156 patients comprising 122 patients with normal perfusion and cardiac function, and remaining 34 patients with heart failure and echocardiographic abnormalities compared the diagnostic performance of four different software program for phase analysis and observed that optimal cutoff value for determining LVD depends on software program used but all software program could be reliably used for phase analysis. Nakajima et al.[12] compared phase parameters among these software's and observed a fair correlation among them but none of the correlation was strong enough to use them interchangeably. In our study, we evaluated the phase analysis parameters in same group of patients using two most commonly used software. A fair correlation between the PSD (r. 54) and HBW (r =0.56) values of both softwares were obtained. In addition, QGS software provides phase parameter LV entropy as a marker of LVD. Entropy increases with worsening of the LV mechanical synchrony.[19] We also found a fair correlation between LV entropy and PSD (r. 44, SEE 5.74), that is, with increase in PSD value there was increase in LV entropy [Figure 1c]. Although it has been shown that the phase parameters may be influenced by myocardial counts, frames per cardiac cycle, dose of radionuclide, stress versus rest study, and gender.[20] The present study has shown that the phase parameters obtained in patients with normal SPECT-MPI study using two different software and their comparison were not interchangeable which could be due to difference in count detection algorithm, etc. We found a fair correlation in various phase analysis parameters in both the software. However, none of the correlation was strong enough to replace the value of other software.

Conclusion

This study was undertaken with the aim to compare the phase analysis parameters obtained in patients with normal SPECT-MPI using two most commonly available software so the derived parameters could be used interchangeably. However, the parameters of stress-only phase analysis in normal gated SPECT MPI studies were dependent on software programs showing different phase variables and the phase analysis results could not be interchangeably used in the same patients. Hence, institutes using gated SPECT-MPI for phase analysis should have their own database for the assessment of LVD.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.