Impact of point spread function modeling and time-of-flight on ...

ORIGINAL ARTICLE Impact of point spread function modeling and time-of-flight on myocardial blood flow and myocardial flow reserve measurements for rubidium-82 cardiac PET Ian S. Armstrong, MSc,a Christine M. Tonge, MSc,a and Parthiban Arumugam, MDa a

Nuclear Medicine, Central Manchester University Hospitals, Manchester, UK

Received Jun 28, 2013; accepted Jan 8, 2014 doi:10.1007/s12350-014-9858-8

Background. Myocardial flow reserve (MFR) obtained from dynamic cardiac positron emission tomography (PET) with rubidium-82 (Rb-82) has been shown to be a useful measurement in assessing coronary artery disease. Advanced PET reconstructions with point spread function modeling and time-of-flight have been shown to improve image quality but also have an impact on kinetic analysis of dynamic data. This study aims to determine the impact of these algorithms on MFR data. Methods. Dynamic Rb-82 cardiac PET images from 37 patients were reconstructed with standard and advanced reconstructions. Area under curve (AUC) of the blood input function (BIF), myocardial blood flow (MBF) and MFR were compared with each reconstruction. Results. No significant differences were seen in MFR for the two reconstructions. A relatively small mean difference in MBF data of 111.9% was observed with advanced reconstruction compared with the standard reconstruction but there was considerable variability in the degree of change (95% confidence intervals of 216.2% to 140.0%). Small systematic relative differences were seen for AUC BIF (mean difference of 26.3%; 95% CI 217.5% to 15.4%). Conclusion: . MFR results from Rb-82 dynamic PET appear to be robust when generated by standard or advanced PET reconstructions. Considerable increases in MBF values may occur with advanced reconstructions, and further work is required to fully understand this. (J Nucl Cardiol 2014;21:467–74.) Key Words: PET/CT imaging Æ coronary blood flow Æ image reconstruction Æ basic science

INTRODUCTION The benefits of myocardial perfusion imaging with rubidium-82 (Rb-82) and positron emission tomography (PET) compared with single photon emission tomography (SPECT) have been widely reported.1,2 A key advantage of Rb-82 cardiac PET is the ability to measure myocardial blood flow (MBF) at rest and stress

Reprint requests: Ian S. Armstrong, MSc, Nuclear Medicine, Central Manchester University Hospitals, Oxford Road, Manchester, UK; [email protected] 1071-3581/$34.00 Copyright Ó 2014 American Society of Nuclear Cardiology.

and, consequently, calculate myocardial flow reserve (MFR)3 with high repeatability and reproducibility across operators and multi-vendor software.4-6 MFR has particular value in the detection of multi-vessel or balanced three-vessel coronary artery disease (CAD)7 and in predicting outcome in patients assessed for ischaemia.8 However, like attenuation-corrected SPECT cardiac imaging, Rb-82 cardiac PET is prone to artifacts due to mis-registration,9 which have been shown to produce errors in MFR measurements.10 Time-of-flight (TOF) PET has been shown to potentially offer increased resilience to mis-registration artifacts.11,12 Furthermore, advanced PET reconstructions with point spread function (PSF) modeling and TOF have been 467

468

I. S. Armstrong et al. Impact of PSF modeling and TOF on MFR

shown, in many studies, to offer considerable benefits in FDG oncology studies.13-15 By contrast, the number of studies that have investigated the impact of these reconstructions on Rb-82 cardiac PET is far smaller.16-18 These studies do show some benefits but the complete impact on all aspects, including kinetic analysis, of Rb82 cardiac PET has yet to be determined. This is relevant because it has been shown, in dynamic brain imaging, that reconstructions with PSF modeling have a significant impact on the image-derived blood input function (BIF).19 The BIF is a key factor in the determination of MBF and changes influence the measurement20 and hence the calculation of MFR. With the assumption that PSF modeling and TOF reconstructions should offer advantages in Rb-82 cardiac PET, in terms of image quality, the rationale of this study was to determine the impact of these algorithms on MBF and MFR measurements in Rb-82 cardiac PET. MATERIALS AND METHODS Patient study group Forty consecutive patients referred for assessment of CAD with Rb-82 cardiac PET were retrospectively selected for inclusion in this study. All data were fully anonymised before analysis. After inspection of dynamic data, three patients were excluded: one for poor myocardial segmentation in the MBF processing software and two for excessive movement during the stress acquisition. The remaining 37 patients consisted of 26 males (mean [range] age: 63.1 year [35–90]; mean [range] weight: 83.4 kg [62–110]); mean [range] body mass index (BMI): 27.9 [17.5–37.6] kg/m2 and 11 females (mean [range] age: 61.6 year [50–82]; mean [range] weight: 93.0 kg [74– 116]); mean [range] BMI: 36.5 kg/m2 [27.9–44.4]. There were 8 male and 8 female patients with obese BMI ([30 kg/m2). In the male patients, 13 were referred with known CAD or previous myocardial infarction (MI), and 13 with suspicion of CAD. One female patient was referred with previous MI with the remaining patients referred with suspicion of CAD. After seeking advice from the institutional research department, ethical approval was not deemed necessary for reprocessing of anonymised retrospective data.

Journal of Nuclear CardiologyÒ May/June 2014

that with a fixed 30-mCi dose, minor detector saturation occurs in less than 1% of all patients.21 Patients underwent pharmacological stress with adenosine using 140 lg/kg/min. The total adenosine infusion lasted 4.5 minutes. For stress imaging, the Rb-82 infusion began 2.5 minutes after the start of the adenosine infusion. All images were acquired on a Siemens Biograph mCT (Siemens Healthcare, Knoxville, US) with extended 21.6-cm axial FOV and 64 slice CT. The scanner uses lutetium oxyorthosilicate (LSO) scintillation crystals and acquires in 3D mode only. For all patients in this study, a single low-dose CT (0.4 mSv) was acquired prior to the rest Rb-82 scan and used for attenuation correction of rest and stress images. A seven minute list mode PET acquisition was started at the same time as the Rb-82 infusion. All scans were checked for mis-registration between PET and CT, using a non attenuation-corrected summed static PET image. If any misregistration was evident, a translational adjustment was applied prior to attenuation-corrected reconstruction. For the dynamic reconstructions, data were reframed as 1 9 10 second, 8 9 5 second, 3 9 10 second, 2 9 20 second, and 4 9 60 second frames. Any adjustments to PET and CT registration were applied to all dynamic frames. Dynamic frames were reconstructed with the department’s standard reconstruction: 3D Ordered Subset Expectation Maximization (OSEM) using 3 iterations, 8 subsets and a 6.5-mm full-width half maximum (FWHM) Gaussian post-filter, which was specified in the manufacturer’s recommendations. Data were also reconstructed with advanced PSF modeling and TOF (PSF?TOF) reconstruction using 2 iterations, 21 subsets and a 6.5-mm FWHM Gaussian post-filter. Given the lack of published data on the use of PSF?TOF reconstruction with Rb-82, the number of iterations and subsets were chosen based upon typical parameters that have been used in recent FDG oncology studies.13 It is noted that the products of iterations and subsets for the two reconstruction algorithms are different but this will illustrate the effect of using the two algorithms with these parameters.

Blood flow analysis Measurements of MBF and MFR were performed using syngoMBF (Siemens Healthcare, Oxford, UK). This software uses a single tissue compartment model for Rb-82 kinetics22 such that tracer concentration in the myocardium over time, Cm(t), is given as

Cm ðtÞ ¼ K1 ek2 t ; Rb-82 cardiac PET imaging All patients were asked to abstain from caffeine for 12 hours before imaging. For both rest and stress imaging, all patients were administered with 1110 MBq (30 mCi) of Rb-82 from a CardiogenÒ Sr-82 generator (Bracco Diagnostics). The rest scan was performed first in all patients. Other studies have used weight-based protocols for determining the level of administered activity5,6 but, due to the requirement of the CardiogenÒ generator to perform a calibration for every change made to the level of administered activity, this was not logistically feasible. Previous work from our institute showed

where K1 and k2 are rate constants for myocardial extraction and clearance, respectively.23 Corrections for myocardial spillover and partial volume effects (PVE) were applied in the software. The BIF was measured using a cylindrical volume of interest, 2-cm diameter and 1.5-cm long, placed in the left ventricular cavity. To measure activity concentration in the myocardium, the myocardium was segmented into 15 rings from apex to base with 36 angular segments in each ring using the automatic edge detection on the software. Activity concentration in each segment was measured using the mean of pixel values within a region of interest of 1-cm radial thickness

Journal of Nuclear CardiologyÒ Volume 21, Number 3;467–74

centered on the middle of the myocardial surface. For each reconstruction, the area under curve (AUC) of the BIF was calculated along with measurements of MBF at rest and stress and MFR (MFR = stress MBF/rest MBF). The ratio of the blood pool activity concentration measurements for the two reconstructions were calculated at each time point in the timeactivity curve (TAC) in order to deduce if any differences occurred at a particular phase of the acquisition. For the BIF AUC, MBF, and MFR data, ratios were obtained between the two reconstructions and compared in males and females and also for patients above and below the mean weight of the study population (85 kg). The decision to use patient weight instead of BMI was a result of previous work that showed the impact of PSF?TOF reconstruction correlated most strongly with patient weight.18

I. S. Armstrong et al. Impact of PSF modeling and TOF on MFR

469

with a mean percentage difference of -6.3% (95% CI -17.5% to ?5.4%), with the Bland-Altman plot shown in Fig. 3. When analyzing the data according to patient gender, significant differences were only seen in the male patients. When comparing the magnitude of reduction of BIF AUC with PSF?TOF in the male patients, BIF AUC was 8% lower in the lower weight group compared with 6% in the upper weight group (P = 0.037). Table 1 shows the BIF AUC obtained from the two reconstruction algorithms. Figure 4 shows the ratios of the BIF AUC obtained from the two reconstructions for the gender and weight groups. Myocardial blood flow and flow reserve

Statistical testing Absolute differences in continuous data did not satisfy Shapiro-Wilk tests for normality (P \ 0.05) and, as such, a non-parametric Wilcoxon’s signed rank test for paired samples was performed to determine differences in BIF AUC, MBF, and MFR. No evidence of non-normality was seen for the relative percentage differences of BIF AUC, MBF, and MFR and, as such, Bland-Altman analysis24 was performed on the relative differences with mean and 95% confidence intervals (95% CI) being calculated. To determine differences between the ratios in patient gender and the two weight groups, a Mann-Whitney U test for unpaired samples was used. Statistical analysis was performed using StatsDirect v2.7.8 (StatsDirect Statistical Software, Altrincham, UK). In all cases, statistical significance was considered for P values less than 0.05.

RESULTS BIF area under curve Figure 1(A) shows the BIF TAC for a single patient at rest and stress. It can be seen from the plot that the structure of the TACs are very similar for rest and stress. Performing the Mann-Whitney test on the BIF AUC for rest against stress data in either reconstruction showed no significant difference between the two data sets. As such, BIF AUC data were analyzed as a single group irrespective of whether they were from the rest or stress acquisitions. The plot of the average ratio of individual TAC points in Fig. 1(B) shows that there is a marked reduction in the BIF activity concentration in the first two frames and then a gradual, but increasing, reduction at the end of the acquisition. Figure 2 shows an example image of the myocardium at stress and rest obtained from the two reconstruction algorithms, demonstrating the location and size of the bloodpool VOI. The BIF AUC was significantly lower (P \ 0.001) with PSF?TOF reconstruction compared with OSEM,

Significant increases (P \ 0.001) were seen in both stress and rest MBF values with PSF?TOF reconstruction compared with OSEM reconstruction. When analyzing the data in the gender and weight categories, all data except for the male patients in the upper weight group showed statistically significant differences, however, even for these obese males, the differences were close to significant (P = 0.074 for stress MBF and P = 0.055 for rest MBF). Table 2 shows the MBF for stress and rest obtained from the two reconstruction algorithms. The mean percentage difference in stress MBF with PSF?TOF was ?10.0% (95% CI -16.5% to ?36.6%) and for the rest MBF, the mean percentage difference was ?13.8% (95% CI -15.8% to ?43.3%). There was no significant difference between the relative percentage differences of MBF ratios at stress and rest indicating a systematic relative increase for PSF?TOF reconstruction and the mean percentage difference for combined stress and rest data was ?11.9% (95% CI -16.2% to ?40.0%). The BlandAltman plots are shown in Fig. 5 for MBF and MFR relative differences. As both stress and rest MBF are increased by approximately the same relative amount, no significant difference was seen in MFR between the two reconstructions. It can be seen in Fig. 5(A) and 5(B) that PSF?TOF reconstruction produced increases of up to 52% for rest MBF and up to 40% for stress MBF. Details from three example patients with high relative increases in MBF are given in Table 3. In order to determine the possible source of the difference, a polar plot was produced to show the ratio of the MBF for the two reconstructions, which is shown in Fig. 6(A). Resting perfusion images from patient #12 are shown in Fig. 6(B). This patient had a small heart with evidence of considerable spill-over of the myocardium into the bloodpool VOI with OSEM reconstruction, which was reduced with PSF?TOF, resulting in a 16% lower BIF AUC with the latter

470

I. S. Armstrong et al. Impact of PSF modeling and TOF on MFR

Journal of Nuclear CardiologyÒ May/June 2014

Figure 1. (A) Example time-activity curve for blood input function at stress (red line) and rest (yellow line). (B) Average ratios (PSF?TOF/OSEM) of the activity concentration measurements at each time point for the two reconstructions. Each data point is obtained by averaging the ratios at a particular time point over all acquisitions.

Figure 2. Screen capture of the myocardial uptake at stress and rest in the last frame of the dynamic reconstruction taken from Syngo MBF analysis for both reconstruction algorithms. The green rectangle represents the blood pool volume of interest.

reconstruction. The second factor evident in the figure is the apparent reduction of septal wall activity in the OSEM images compared with PSF?TOF images. In this region, an increase of MBF by approximately 60– 70% was observed with PSF?TOF as can be seen in Fig. 6(A). The increase in global resting MBF for this patient with PSF?TOF reconstruction was 35%. In all cases, no significant differences in the relative percentage differences were seen for the weight and gender groups.

Figure 3. Bland-Altman plots of the relative percentage differences for BIF AUC data for the two reconstructions. The solid line shows the mean percentage difference between reconstructions while the dashed lines show 95% confidence intervals.

DISCUSSION We believe that this is only the second study to investigate the use of advanced PET reconstructions with PSF modeling and TOF in Rb-82 cardiac PET and the first to evaluate their impact on MBF and MFR for dynamic data. The work has shown that MFR measurements appear to be robust for the two reconstruction algorithms. However, we noted a considerable variability in the relative changes seen for MBF between the two reconstructions (Fig. 5(A) and

Journal of Nuclear CardiologyÒ Volume 21, Number 3;467–74

I. S. Armstrong et al. Impact of PSF modeling and TOF on MFR

471

Table 1. Data for area under curve (AUC) of the blood input function (BIF) obtained from the two reconstructions

Median for BIF AUC (Bq/mL 3 seconds)

All patients \85 kg [85 kg Male \85 kg [85 kg Female \85 kg [85 kg

No. images

OSEM (3106)

PSF 1 TOF (3106)

P

74 40 34 52 34 18 22 6 16

20.5 22.9 18.5 22.0 23.2 19.3 17.5 18.7 17.2

19.3 21.2 18.1 20.4 21.4 18.2 17.9 18.9 17.3

\0.001 \0.001 \0.001 \0.001 \0.001 \0.001 0.330 0.438 0.720

Figure 4. Box-whisker plots of the ratios between the two reconstructions (PSF?TOF/OSEM) for the area under curve (AUC) of the blood input function (BIF) in the male (M) and female (F) upper and lower weight groups. Statistical testing was performed using the Mann–Whitney test.

5(B)). In addition, substantial increases in MBF, of up to 52%, were seen in some patients. This is most likely to be a consequence of increased cavity contrast, which has been reported with PSF reconstruction.15 This reduces the amount of spillover of myocardial activity into the bloodpool VOI, particularly in the later frames, as shown in Figs. 1(B) and 2, resulting in an apparent reduction in the BIF and hence greater MBF. This appeared to be most prominent in small hearts where the bloodpool VOI occupied the majority of the ventricular cavity. In addition, a reduction of the BIF was also seen in the first and second frames (Fig. 1(B)) with PSF?TOF. This appeared to be due to a reduction of spillover from activity in the right ventricle with PSF?TOF compared with OSEM

because of the increased convergence of PSF?TOF as a consequence of the greater product of iterations and subsets. In some patients, a regional increase in perfusion was seen with PSF?TOF resulting in a greater regional MBF that lead to an increased global MBF with PSF?TOF. This may be due to a reduction in PVE when using PSF?TOF reconstruction in a myocardium that is particularly thin. Unfortunately, the lack of contrast enhancement in the CT data in this study meant that it was not possible to visualize the myocardial wall from the cavity in order to estimate the thickness of the myocardium. One may consider that a reduction in PVE would produce more accurate MBF and this provides scope for follow-up work to include anatomical measurements of myocardial thickness. However, partial volume corrections and the nonlinear extract correction of rubidium3 that are incorporated into dynamic processing software are likely to require modification to account for differences in PVE from the advanced reconstructions to maintain consistent flow results. The changes seen in this study to MBF results may be clinically significant and caution should be applied if processing software is not adapted for data using advanced reconstructions. However, the MFR values, which are considered the prime prognostic indicator, both in predicting three-vessel disease7 and patient outcome,8 do not appear to be affected by the advanced reconstruction. The magnitude of the relative increase for the MBF data was not found to be dependent on gender or weight. Conversely, the relative differences for BIF AUC obtained from PSF?TOF and OSEM did appear to be dependent on weight in the male patients. It is not clear why these differences only occur in the male patients. A possible explanation for why the relative increases of

472

I. S. Armstrong et al. Impact of PSF modeling and TOF on MFR

Journal of Nuclear CardiologyÒ May/June 2014

Table 2. Myocardial blood flow values for the two reconstructions at stress and rest

Median stress MBF (mL/g/min)

All \85 kg [85 kg Male \85 kg [85 kg Female \85 kg [85 kg

n

OSEM

PSF 1 TOF

37 20 34 26 17 9 11 3 8

2.61 2.89 2.49 2.55 3.08 2.10 2.65 2.24 2.74

2.80 2.99 2.63 2.66 3.06 2.45 2.89 2.57 2.94

Median rest MBF (mL/g/min) P \0.001 0.019 0.005 0.008 0.045 0.074 0.014 – 0.039

OSEM

PSF 1 TOF

1.21 1.30 1.11 1.23 1.34 0.83 1.15 1.06 1.18

1.27 1.46 1.10 1.28 1.47 1.08 1.23 1.14 1.32

P \0.001 0.006 0.002 0.002 0.020 0.055 0.005 – 0.020

Statistically testing was performed using the Wilcoxon signed rank test. A P value could not be deduced for female patients in the upper weight category due to the sample size

Figure 5. Bland-Altman plots of the relative percentage differences for rest MBF (A), stress MBF (B), and MFR (C) for the two reconstructions. The solid line shows the mean percentage difference between reconstructions while the dashed lines show the 95% confidence limits.

Table 3. Three example patients that were considered to have a large percentage increase in MBF with PSF ? TOF reconstruction compared with OSEM

Patient #12 (M, 63 kg) #24 (M, 103 kg) #27 (F, 94 kg)

Stress MBF (mL/g/min)

Rest MBF (mL/g/min)

OSEM

PSF 1 TOF

OSEM

PSF 1 TOF

OSEM

PSF 1 TOF

3.22 2.58 3.16

4.10 (?27%) 3.60 (?40%) 4.02 (?27%)

1.48 0.84 1.21

1.99 (?35%) 1.28 (?52%) 1.65 (?36%)

2.20 3.12 2.62

2.10 (-4.5%) 2.92 (-5.3%) 2.48 (-6.4%)

MBF do not appear to be dependent on gender or weight is that any effects of the PSF?TOF reconstruction that are related to patient gender or weight are likely to be similar for the measurements of the

MFR

BIF and also the Rb-82 activity in the myocardium and hence cancel out during the MBF calculation. On our scanner, the use of PSF modeling and TOF reconstruction results in approximately 50% longer

Journal of Nuclear CardiologyÒ Volume 21, Number 3;467–74

I. S. Armstrong et al. Impact of PSF modeling and TOF on MFR

473

Figure 6. (A) Ratio of resting MBF (PSF?TOF/OSEM) for patient #12, which had a global increase of 35% for their MBF with PSF?TOF compared with OSEM reconstruction. The two perfusion images in (B) show a reduction in perfusion in the septal wall with OSEM compared with PSF?TOF that correlated with the area of increased MBF. The window levels are set equal in the two uptake images. The flow data for the patient are given in Table 3.

reconstruction times, which may impact on logistics during a busy PET service. However, TOF may offer potential advantages in reducing attenuation artifacts from mis-registration in cardiac studies,11 which have been shown to impact detrimentally on MBF and MFR.9 From our experience, excellent motion-free dynamic images can be obtained in almost all resting studies through effective communication and patient preparation. Despite the best efforts, motion can occur during the stress acquisition due to patient discomfort from the effects of the adenosine infusion. While registration is checked, and adjusted if necessary using the summed static stress perfusion images, there is currently no way of correcting for motion-related misregistration on a frame-by-frame basis on the dynamic images. Motion may cause mis-registration in some of the dynamic frames between the stress PET and the CT for attenuation correction and hence errors in MBF and MFR results. As such, the use of a reconstruction algorithm that reduces the impact of errors from misregistration would be of benefit. The findings from this study and other work would encourage the use of TOF reconstruction, which may reduce such errors without impacting on the results obtained for MFR. It is our intention to evaluate this hypothesis in future work. Study limitations This study has limitations. Statistically significant differences were seen for BIF AUC and MBF for the two reconstructions in this study. However, without a gold standard measurement, it is impossible to deduce with certainty which reconstruction resulted in the more

accurate measurements. While we note that considerable differences in MBF can occur in small hearts due to reductions in spillover when using PSF?TOF reconstruction, we are unable to draw solid conclusions from apparent changes in myocardial uptake that may be due to PVE. The lack of information on the myocardial wall visualization in this study does not allow for estimates of myocardial thickness to be estimated, which would provide evidence to support the PVE hypothesis. There is a variety of possible explanations for the differences seen with patient gender and weight in MBF that could include: differences in reconstruction convergence at different patient weights and differences in size of the myocardium, which would impact on myocardial spillover. Due to relatively small sample size in this study, the exact reasoning cannot be deduced confidently. CONCLUSION This preliminary study has indicated that myocardial flow reserve measurements from dynamic rubidium82 dynamic PET appear to be robust when generated by standard or advanced PET reconstruction algorithms with PSF modeling and TOF. Differences in MBF values were seen, possibly due to reductions in partial volume effects, which warrant further investigation, supported by additional anatomical imaging such as contrast-enhanced CT or cardiac MR. Potential improvements in the robustness of the technique when using PSF modeling and TOF reconstruction, such as reduced sensitivity to mis-registration, encourage the adoption of these advanced algorithms into routine use for rubidium-82 cardiac PET.

474

I. S. Armstrong et al. Impact of PSF modeling and TOF on MFR

References 1. Bateman TM, Heller GV, McGhie AI, et al. Diagnostic Accuracy of Rest/Stress ECG-Gated Rb-82 Myocardial Perfusion PET: Comparison with ECG-Gated Tc-99m sestamibi SPECT. J Nucl Cardiol 2006;13(1):24–33. 2. Flotats A, Bravo P, Fukushima K, Chaudhry M, Merrill J, Bengel F. 82 Rb PET Myocardial Perfusion Imaging is Superior 99mTc-Labelled Agent SPECT in Patients with Known or Suspected Coronary Artery Disease. Eur J Nucl Med Mol Imaging 2012;39(8):1233–9. 3. Prior J, Allenbach G, Valenta I, et al. Quantification of Myocardial Blood Flow with 82Rb Positron Emission Tomography: Clinical Validation with 15O-Water. Eur J Nucl Med Mol Imaging 2012;39(6):1037–47. 4. Manabe O, Yoshinaga K, Katoh C, Naya M, deKemp RA, Tamaki N. Repeatability of Rest and Hyperemic Myocardial Blood Flow Measurements with 82Rb Dynamic PET. J Nucl Med 2009;50(1):68–71. 5. Efseaff M, Klein R, Ziadi M, Beanlands R, deKemp R. ShortTerm Repeatability of Resting Myocardial Blood Flow Measurements Using Rubidium-82 PET Imaging. J Nucl Cardiol 2012;19(5):997–1006. 6. deKemp RA, Declerck J, Klein R, et al. Multisoftware Reproducibility Study of Stress and Rest Myocardial Blood Flow Assessed with 3D Dynamic PET/CT and a 1-Tissue-Compartment Model of 82Rb Kinetics. J Nucl Med 2013;54(4):571–7. 7. Ziadi MC, deKemp RA, Williams MSK, et al. Does Quantification of Myocardial Flow Reserve Using Rubidium-82 Positron Emission Tomography Facilitate Detection of Multivessel Coronary Artery Disease? J Nucl Cardiol 2012;19(4):670–80. 8. Ziadi MC, deKemp RA, Williams KA, Guo A, Chow BJW, Renaud JM, et al. Impaired Myocardial Flow Reserve on Rubidium82 Positron Emission Tomography Imaging Predicts Adverse Outcomes in Patients Assessed for Myocardial Ischemia. J Am Coll Cardiol 2011;58:740–8. 9. Loghin C, Sdringola S, Gould KL. Common Artifacts in PET Myocardial Perfusion Images Due to Attenuation–Emission Misregistration: Clinical Significance, Causes, and Solutions. J Nucl Med 2004;45(6):1029–39. 10. Rajaram M, Tahari AK, Lee AH, et al. Cardiac PET/CT Misregistration Causes Significant Changes in Estimated Myocardial Blood Flow. J Nucl Med 2013;54(1):50–4. 11. Conti M. Why is TOF PET reconstruction a more robust method in the presence of inconsistent data? Phys Med Biol 2011;56(1):155–68.

Journal of Nuclear CardiologyÒ May/June 2014

12. Armstrong I, Tout D, Tonge C, Arumugam P. Time-of-Flight Reduces the Severity of CT Mis-registration Artefacts in Rubidium-82 Cardiac PET [Abstract]. J Nucl Med 2013;54(Suppl 2):1636. 13. Akamatsu G, Ishikawa K, Mitsumoto K, et al. Improvement in PET/CT Image Quality with a Combination of Point-Spread Function and Time-of-Flight in Relation to Reconstruction Parameters. J Nucl Med 2012;53(11):1716–22. 14. Andersen FL, Klausen TL, Loft A, Beyer T, Holm S. Clinical Evaluation of PET Image Reconstruction Using a Spatial Resolution Model. Eur J Radiol 2013;82:862–9. 15. Karp JS, Surti S, Daube-Witherspoon ME, Muehllehner G. Benefit of Time-of-Flight in PET: Experimental and Clinical Results. J Nucl Med 2008;49:462–70. 16. LeMeunier L, Slomka PJ, Dey D, et al. Enhanced Definition PET for Cardiac Imaging. J Nucl Cardiol 2010;17(3):414–26. 17. DiFilippo F, Brunken R. Benefit of Time-of-Flight Reconstruction for Cardiac PET of Obese Patients [Abstract]. J Nucl Med 2013;54(Suppl 2):405. 18. Armstrong I, Tonge CM, Arumugam P. The Impact of Advanced Reconstruction on Myocardial Image Noise in Rubidium Myocardial Perfusion PET. J Nucl Cardiol 2013;20(Suppl 1):S23. 19. Lewis J, Anton-Rodriguez J, Carter SF, Herholz K, Asselin MC, Hinz R (2012) Optimization of high resolution PET iterative reconstruction with resolution modeling for image derived input function. In: IEEE Nuclear Science Symposium Conference Record, pp 3999–4000. 20. Vasquez AF, Johnson NP, Gould KL. Variation in Quantitative Myocardial Perfusion due to Arterial Input Selection. JACC Cardiovasc Imaging 2013;6(5):559–68. 21. Tout D, Tonge CM, Muthu S, Arumugam P. Assessment of a Protocol for Routine Simultaneous Myocardial Blood Flow Measurement and Standard Myocardial Perfusion Imaging with Rb-82 on a High Count Rate PET System. Nucl Med Commun 2012;33(11):1202–11. 22. Coxson PG, Huesman RH, Borland L. Consequences of Using a Simplified Kinetic Model for Dynamic PET Data. J Nucl Med 1997;38(4):660–7. 23. Lortie M, Beanlands R, Yoshinaga K, Klein R, DaSilva J, deKemp R. Quantification of Myocardial Blood Flow with 82Rb Dynamic PET Imaging. Eur J Nucl Med Mol Imaging 2007;34(11):1765–74. 24. Bland JM, Altman DG. Measuring Agreement in Method Comparison Studies. Stat. Methods Med. Res 1999;8:135–60.