ORIGINAL ARTICLE Interpolated average CT for cardiac PET/CT attenuation correction Greta S. P. Mok, PhD,a Cobie Y. T. Ho, BSc,a Bang-Hung Yang, PhD,b,c and Tung-Hsin Wu, PhDb a
Biomedical Imaging Laboratory, Department of Electrical and Computer Engineering, Faculty of Science and Technology, University of Macau, Macau, SAR, China b Department of Biomedical Imaging and Radiological Sciences, National Yang Ming University, Taipei, Taiwan c Department of Nuclear Medicine, Taipei Veterans General Hospital, Taipei, Taiwan Received Feb 12, 2015; accepted Apr 1, 2015 doi:10.1007/s12350-015-0140-5
Background. Previously, we proposed interpolated averaged CT (IACT) for improved attenuation correction (AC) in thoracic PET/CT. This study aims to evaluate its feasibility and effectiveness on cardiac PET/CT. Methods. We simulated 18F-FDG distribution using the XCAT phantom with normal and abnormal cardiac uptake. Average activity and attenuation maps represented static PET and respiration average CT (ACT), respectively, while the attenuation maps of end-inspiration/expiration represented 2 helical CTs (HCT). IACT was obtained by averaging the 2 extreme phases and the interpolated phases generated between them. Later, we recruited 4 patients who were scanned 1 hr post 315-428 MBq 18F-FDG injection. Simulated and clinical PET sinograms were reconstructed with AC using (1) HCT, (2) IACT, and (3) ACT. Polar plots and the 17-segment plots were analyzed. Two regions-of-interest were drawn on lesion and background area to obtain the intensity ratio (IR). Results. Polar plots of PETIACT-AC were more similar to PETACT-AC in both simulation and clinical data. Artifacts were observed in various segments in PETHCT-AC. IR differences of HCT as compared to the phantom were up to *20%. Conclusions. IACT-AC reduced respiratory artifacts and improved PET/CT matching similarly to ACT-AC. It is a promising low-dose alternate of ACT for cardiac PET/CT. (J Nucl Cardiol 2015) Key Words: PET/CT Æ cardiac imaging Æ attenuation correction Æ respiratory artifacts
INTRODUCTION The power of integrated PET/CT imaging has been well recognized for oncological applications, and its effectiveness in cardiology is a growing interest in both research and clinical practice. Common cardiac PET 82 perfusion agents include 13NH3, H15 Rb while 2 O, and
Reprint requests: Greta S. P. Mok, Ph. D, Biomedical Imaging Laboratory, Department of Electrical and Computer Engineering, Faculty of Science and Technology, University of Macau, Avenida da Universidade, Taipa, Macau, SAR, China; [email protected], [email protected] 1071-3581/$34.00 Copyright 2015 American Society of Nuclear Cardiology.
the most common tracer for viability assessment is 18FFDG.1 Although SPECT is still the primary imaging modality for myocardial perfusion imaging (MPI), PET emerges as an increasingly popular alternate to MPI SPECT, providing superior image quality as well as diagnostic accuracy especially for obese patients, patients undergoing pharmacologic stress, and equivocal cases.2-6 In addition, quantitative analyses such as myocardial blood flow (MBF), coronary flow reserve (CFR), and left ventricular ejection fraction (LVEF) are feasible on PET/CT, with improved risk stratification and patient management.7,8 The use of CT in PET/CT provides intrinsically registered anatomical information for PET functional
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
images and improves the quantitative accuracy with CTbased attenuation correction (AC). For cardiac imaging, added benefits from CT include calcium scoring9 and coronary CT angiogram (CTA)10 from high-end CT scanners. However, due to the potential mis-registration of the sequential PET and CT scans where the acquisition time of PET spans over multiple respiratory cycles, but conventional helical CT (HCT) scan just spans over few seconds, using the mis-registered CT for PET AC will result in respiratory misalignments and artifacts. While signature curvilinear artifacts are often seen in the lung-diaphragm border, extra cardiac misalignments are sometimes observed in the tissues around the lung and left-ventricle interface, where PET myocardial uptake can be overlaid on the lung tissues on the CT images.11 Both effects can potentially affect the further quantitative cardiac analysis. For MPI PET/CT, more than 40% of the studies have artifactual defects in the cardiac region when no interventions are taken to address the PET/CT misalignment, causing significant diagnostic errors.11 Different types of mismatch lead to artifacts and increases in myocardial non-uniformity.12,13 Mis-registration of the stress PET/CT also affect the global and regional myocardial blood flow estimation.14,15 Besides simple manual or automatic registration based on the outline of the heart for the misaligned PET and CT as suggested by Martinez-Mo¨ller et al,14 breathing instruction, CT protocols, and gated 4-dimensional (4D) PET/CT have been investigated mostly to reduce the PET/CT misalignments and artifacts so far. A detailed summary for different motion artifact reduction methods can be found from our previous review article.16 Among them, respiration average CT (ACT) which acquired 4D CT in the thoracic region was developed for AC in PET and showed significantly less misalignments and artifacts as compared with conventional HCT-AC.17,18 Alessio et al further evaluated both average and intensity maximum images of 4D CT and
Journal of Nuclear Cardiology
indicated the later method had better alignments between PET and CT.19 The main problem of ACT is relatively high radiation dose. Previously, we developed an interpolated average CT (IACT) method as a low-dose alternate to ACT for AC to reduce PET/CT misalignments. This method requires the aid of an active breathing controller (ABC) to acquire the end-inspiration and end-expiration phases of the patients. Its effectiveness in image quality and quantitative accuracy improvement for thoracic lesions were demonstrated in our previous simulation20 and clinical studies.21,22 In this study, we aim to evaluate the potential of using IACT for PET AC in cardiology applications. Both simulation and patient data were analyzed to compare the performance of IACT, ACT, and HCT.
MATERIALS AND METHODS Simulation Study We simulated a male patient injected with 18F-FDG distribution using the digital 4D Extended Cardiac Torso phantom (XCAT)23 with cardiac motion and respiratory motion of 2, 3, and 4 cm (Figure 1a). Normal cardiac activity uptake as well as an ischemic heart was modeled, where a longitudinal 2-cm and circumferential 60 lesion with 60% of the normal wall uptake was placed at the lateral and inferior wall of the left ventricle (LV) myocardium individually (Figure 1b, c). The respiratory period of 5.9 seconds was binned into 13 phases, with average activity and attenuation maps to represent static PET and ACT, respectively. The attenuation maps of end-inspiration and end-expiration represented 2 different HCTs (HCT-1 and HCT-8). We used B-spline method to calculate the deformation vectors which include lateral, anterior-posterior, and inferiorsuperior displacement for each voxel between HCT-1 and HCT-8 based on the Insight Segmentation and Registration Toolkit (ITK).24 The forward deformation vector, i.e., from HCT-1 to HCT-8, and backward deformation vector, i.e., from
Figure 1. (A) Average 18F-FDG activity map with cardiac motion and respiratory motion of 2 cm. Short-axis images showing lesion placed at (B) lateral wall and (C) inferior wall.
Journal of Nuclear Cardiology
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
Figure 2. (A) Respiratory phase#1 of the attenuation map represented HCT obtained at end-inspiration (HCT-1); (B) respiratory phase#8 of the attenuation map represented HCT obtained at end-expiration (HCT-8); (C) average attenuation map represented ACT, (D) IACT.
Figure 3. Polar plots generated from the noise-free PET short-axis reconstructed images for a normal heart simulation with respiratory motion amplitude of (A) 2 cm; (B) 3 cm; (C) 4 cm, and for a heart with a lesion placed at (D) inferior wall and (E) lateral wall with respiratory motion of 2 cm. The ROIs drawn for calculating IR were marked in the red squares. HCT-8 to HCT-1, were then used to generate 10 interpolated phases using an empirical diaphragmatic movement function.25 The IACT was obtained by averaging the original extreme
phases, i.e., HCT-1 and HCT-8, and the interpolated phases. More details about the IACT generation can be found in our previous work.21
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
To simulate noisy CT data, CT projections were generated from different AC maps using an analytical projector and then added with Gaussian noise. The projections were then reconstructed with filtered back-projection method to get the reconstructed CT images with similar noise level as in the clinical CT data for further AC in PET (Figure 2), where the noise measurement was based on the standard deviation of a region-of-interest (ROI) drawn in a homogenous liver region on the clinical CT patient data.22 Noise-free and noisy sinograms with attenuation modeling were generated for 2 bed positions to cover the thoracic region and reconstructed with different AC maps (ACT, HCTs, and IACTs) by Software for Tomographic Image Reconstruction (STIR),26 using ordered-subset expectation maximization (OS-EM) method with up to 300 updates. Noisy PET sinogram was generated by adding Poisson noise on the noise-free data based on a count level of *9.5 M, representing a 20 minutes whole-body PET acquisition. The simulation geometry was based on the GE Discovery STE PET scanner.
Patient Study Between October 2012 and December 2012, we recruited 4 male patients with age from 41 to 86 who had their scheduled
Journal of Nuclear Cardiology
whole-body PET/CT scans under the ethics approval of the Institutional Review Board of Taipei Veterans General Hospital. Written informed consents were obtained from all patients. They were scanned 1 hour post 315 to 428 MBq 18F-FDG injection on a clinical PET/CT scanner (Discovery VCT, GE Medical Systems, Milwaukee, WI), which operated in 3dimensional (3D) mode with transaxial field-of-views (FOVs) of 70 and 50 cm for PET and CT, respectively. During the whole-body PET scans, 7 bed positions with 3 minutes/bed were acquired for each patient, and PET sinograms were reconstructed with AC using (1) standard shallow free breathing HCT (120 kV, smart mA (30-150 mA), 0.984:1 pitch); (2) IACT generated from 2 end-inspiration and end-expiration breath-hold HCTs (120 kV, 10 mA) aided by an ABC prototype; and (3) ACT (120 kV, 10 mA, 5.9 seconds duration). The home-made ABC system integrates a spirometer, an air mask, a tube-valve system, and a C??-based acquisition program to ensure the patients could hold their breaths at the desired end-inspiration and end-expiration phases, with breathing amplitude similar to the free breathing state.21,27 The IACT was then generated using the same procedure as mentioned in the simulation with the captured extreme CT phases. All data were reconstructed on the scanner workstation using 3D OSEM method with 2 iterations and 28 subsets. Besides AC, the
Figure 4. (A) The PET short-axis reconstructed images with AC using different CT protocols for patient#1; (B) the corresponding polar plots; (C) polar plots for patient#2.
RESULTS The polar plots of PETIACT-AC were more similar to PETACT-AC from visual assessment in both simulations (Figure 3) and clinical data (Figure 4). Artifacts were observed in PETHCT-AC in various locations including anterior, anterolateral, anteroseptal, and inferior segments of LV. In simulations, the IR differences were close for PETIACT-AC and PETACT-AC, while it can reach *20% for PETHCT-8-AC (Table 1). The seventeen-segment plots of simulations and clinical data indicating the RD as compared to the reference are shown in Figures 5 and 6. The RD was substantially higher with more nonuniformity observed for HCT-AC. For the PET/CT fusion images, significant misalignment was observed at the interface between the lung and left-ventricle in PETHCT-AC/HCT (Figure 7). DISCUSSION The mismatch of emission data and CT-based transmission map was a well-recognized problem, no matter for gated or static emission acquisition. Respiratory gating in PET reduces substantial motion blurring and phase matched CT from 4D or virtual 4D CT was proposed to further improve the PET-CT
0.85 0.88 0.86 0.92 0.98 1.01 0.95 1.08 0.96 0.98 0.93 1.06 3.86 3.87 3.79 3.88 3.47 3.50 3.33 3.59 3.48 3.52 3.26 3.63 3.28 3.41 3.25 3.57 3.42 3.53 3.17 3.87 3.35 3.46 3.03 3.87 *%diff = (IRrecon - IRphantom)/IRphantom 9 100%
The clinical PET reconstructed images with AC using different AC maps, i.e., PETHCT-AC, PETACT-AC, and PETIACT-AC were also registered with associated CT for further visual assessment.
4 cm
ð2Þ
3 cm
ROIlesion : ROIbg
Table 1. Intensity ratios (IRs) for different AC schemes and motion amplitudes
IR ¼
2.83 3.28 -0.59 5.26 7.02 7.04 4.00 9.11 11.98 12.49 11.36 15.48
The original phantom polar plot was used as the reference in simulation, while the polar plot generated from PETACT-AC was used as the reference in patient study respectively. In simulation, two ROIs were drawn in lesion (ROIlesion) and background (ROIbg) area in the polar plots to calculate the intensity ratio (IR, Figure 3d, e):
0.81 0.81 0.78 0.83 0.88 0.88 0.86 0.90 0.78 0.79 0.78 0.81
ð1Þ
3.86 3.87 3.79 3.87 3.47 3.50 3.33 3.59 3.48 3.52 3.26 3.63
I recon Iref 100 %: RD ¼ Iref
3.12 3.14 2.96 3.21 3.07 3.09 2.85 3.23 2.72 2.77 2.53 2.93
Polar plots were generated from the reconstructed shortaxis images for both simulations and patient study. Based on the American Heart Association guidelines, seventeen segments were drawn in the polar plots.28 Relative difference (RD) of the average count value of each segment on the reconstruction images using different AC methods (Irecon) and reference images (Iref) was computed to quantify the recovery of the activity concentration:
ACT IACT HCT-1 HCT-8 ACT IACT HCT-1 HCT-8 ACT IACT HCT-1 HCT-8
Data Analysis
2 cm
PET sinograms were corrected for random, scatter, and isotope decay.
-0.09 3.43 0.88 8.23 5.30 8.16 1.96 15.31 8.22 10.25 4.41 19.56
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
Motion AC amplitude scheme ROIlesion_inferior ROIbg_inferior IRinferior %diffinferior* ROIlesion_lateral ROIbg_lateral IRlateral %difflateral*
Journal of Nuclear Cardiology
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
Journal of Nuclear Cardiology
Figure 5. Seventeen-segment analysis based on the noisy polar plots for a normal heart simulation with motion amplitude of (A) 2 cm; B 3 cm and (C) 4 cm. The RD as compared to the phantom is displayed as the grayscale in each segment, with dark red indicates 40% and light yellow indicates 0%.
alignment.29,30 Different PET respiratory phases can also be transformed and registered to a reference CT phase for further reconstruction.31 However, respiratory gating was still not a clinical routine procedure probably due to implementation complexities and the increased noise level in each respiratory bin. While ACT was proven to be an effective AC method for respiration-ungated cardiac PET,32 similar radiation concern arises as 4D CT. Thus, previously we proposed IACT method as an ACT alternate with lower radiation dose. In the clinical study, the estimated mean effective doses were 0.38, 2.26, and 2.01 mSv for IACT, HCT, and ACT, respectively. The IACT reduced 83% and 81% effective dose as compared to HCT and ACT. The robustness of the ABC device was discussed in our previous work.21 Our recruited patients all showed good compliance to the device, with minimal increase of the radiation dose as the lowest tube current, i.e., 10 mA, was used in the study. The results of IACT substantially approached to those of the ACT as compared to HCT
(Figures 4 and 6), indicating the effectiveness of our algorithm and the ABC. Previously patients’ voluntarily breath-hold was used to capture the 2 extreme phases for generating IACT, without using ABC.33 However, we found that the result was highly unpredictable and misalignments still persisted or even exaggerated for some cases. Thus, the ABC device is an essential component for IACT-AC. In the simulation study, we found that the IR performance of HCT-1, i.e., HCT acquired at endinspiration phase, was superior to HCT-8, i.e., HCT acquired at end-expiration phase and even ACT. That might due to the fact that the diaphragm was at a lower position for end-inspiration state and would cause less attenuation to the heart. Less blurring from the HCT might also improve the lesion delectability. However, from the global assessment as in the 17-segment plot, HCT-1 was superior to HCT-8 but was still inferior to both ACT and IACT. The general matching of static PET and CT was markedly improved for the average CT techniques (Figure 7).
Journal of Nuclear Cardiology
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
Figure 6. Seventeen-segment analysis based on the clinical data as shown in Figure 4, with (A) & (B) generated from patient#1 and (C) & (D) generated from patient#2. The RD as compared to the PETACT-AC is displayed as the grayscale in each segment, with dark red indicates 40% and light yellow indicates 0%.
Figure 7. Fusion images of (A) PETACT-AC /ACT; (B) PETIACT-AC /IACT; and (C) PETHCT-AC /HCT for patient#1.
In this study, only static 18F-FDG PET scan was evaluated. The ECG gating 18F-FDG scan is also a common clinical procedure, using a single breath-hold or shallow free breathing HCT as the attenuation map for different PET phases. Further evaluation in using ACT/IACT for cardiac gated PET scan is warranted. Interpolated CT (ICT), a precursor of IACT, might be more suitable for AC in respiratory gated PET scan and is currently under investigation by our group. More evaluations on using IACT-AC on myocardial perfusion PET are also warranted.
CONCLUSION Our simulation and clinical studies accordantly showed that the respiratory artifacts can be reduced, while the uniformity and the PET/CT matching can be improved with ACT/IACT-AC in cardiac PET, as compared to standard HCT-AC. The IACT is a good low-dose alternate of ACT and has the potential to improve cardiac PET/CT imaging. Acknowledgments
NEW KNOWLEDGE GAINED
• HCT-AC leads to artifacts in anterior, anterolateral, anteroseptal, and inferior segments of LV, degraded cardiac uniformity, and lesion quantitation on PET cardiac images. • Respiratory average CT, i.e., ACT and IACT, can reduce the PET/CT image mismatch, reduce artifacts, and improve quantitation for cardiac PET, according to the polar plots and 17-segment analysis. • Performances of IACT and ACT are similar for cardiac PET, while IACT shows lower estimated dose.
The authors would like to thank Mr. Tao Sun from Nuclear Medicine and Medical Imaging Research Center at K.U. Leuven and Ms Nien-Yun Wu from the Department of Nuclear Medicine at Taipei Veterans General Hospital for their technical and experimental assistance. This work was supported by research grants (MYRG185(Y4-L3)-FST11-MSP and MYRG077(Y3-L2)-FST12-MSP) from University of Macau, Macau, and a research grant (MOST 103-2314-B195-001-MY3) from the Ministry of Science and Technology of Taiwan, Taiwan.
Disclosure None.
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
References 1. Matsunari I, Taki J, Nakajima K, Tonami N, Hisada K. Myocardial viability assessment using nuclear imaging. Ann Nucl Med 2003;17:169-79. 2. Heller GV, Calnon D, Dorbala S. Recent advances in cardiac PET and PET/CT myocardial perfusion imaging. J Nucl Cardiol 2009;16:962-9. 3. Di Carli MF, Dorbala S, Meserve J, El Fakhri G, Sitek A, Moore SC. Clinical myocardial perfusion PET/CT. J Nucl Med 2007;48:783-93. 4. Knaapen P, de Haan S, Hoekstra OS, Halbmeijer R, Appelman YE, Groothuis JGJ, et al. Cardiac PET-CT: advanced hybrid imaging for the detection of coronary artery disease. Neth Heart J 2010;18:90-8. 5. Bateman TM, McGhie AI, O’Keefe JH, Thompson RC, Moutray KL, Bryngelson JR, et al. High clinical value of follow-up myocardial perfusion PET in patients with a diagnostically indeterminate myocardial perfusion SPECT study. Circulation 2003;108:454. 6. Bateman TM, Heller GV, McGhie AI, Friedman JD, Case JA, Bryngelson JR, 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;13:24-33. 7. Shah AR, Lertsburapa K, Ahlberg A, Katten D, Volker L, Cullom SJ, et al. What is the prognostic value of myocardial perfusion imaging using rubidium-82 positron emission tomography in obese patients with known or suspected CAD? J Am Coll Cardiol 2008;51:A173-A. 8. Yoshinaga K, Chow BJW, Williams K, Chen L, Dekemp RA, Garrard L, et al. What is the prognostic value of myocardial perfusion imaging using rubidium-82 positron emission tomography? J Am Coll Cardiol 2006;48:1029-39. 9. Burkhard N, Herzog BA, Husmann L, Pazhenkottil AP, Burger IA, Buechel RR, et al. Coronary calcium score scans for attenuation correction of quantitative PET/CT N-13-ammonia myocardial perfusion imaging. Eur J Nucl Med Mol I 2010;37:517-21. 10. Hacker M, Jakobs T, Matthiesen F, Vollmar C, Nikolaou K, Becker C, et al. Comparison of spiral multidetector CT angiography and myocardial perfusion imaging in the noninvasive detection of functionally relevant coronary artery lesions: First clinical experiences. J Nucl Med 2005;46:1294-300. 11. Gould KL, Pan T, Loghin C, Johnson NP, Guha A, Sdringola S. Frequent diagnostic errors in cardiac PET/CT due to misregistration of CT attenuation and emission PET images: A definitive analysis of causes, consequences, and corrections. J Nucl Med 2007;48:1112-21. 12. McQuaid S, Hutton B. Sources of attenuation-correction artefacts in cardiac PET/CT and SPECT/CT. Eur J Nucl Med Mol Imaging 2008;35:1117-23. 13. 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;45:1029-39. 14. Martinez-Mo¨ller A, Souvatzoglou M, Navab N, Schwaiger M, Nekolla SG. Artifacts from Misaligned CT in Cardiac Perfusion PET/CT Studies: Frequency, Effects, and Potential Solutions. J Nucl Med 2007;48:188-93. 15. Rajaram M, Tahari AK, Lee AH, Lodge MA, Tsui B, Nekolla S, et al. Cardiac PET/CT misregistration causes significant changes in estimated myocardial blood flow. J Nucl Med 2013;54:50-4.
Journal of Nuclear Cardiology
16. Sun T, Mok GS. Techniques for respiration-induced artifacts reductions in thoracic PET/CT. Quant Imaging Med Surg 2012;2:4652. 17. Pan T, Mawlawi O, Nehmeh SA, Erdi YE, Luo D, Liu HH, et al. Attenuation Correction of PET Images with Respiration-Averaged CT Images in PET/CT. J Nucl Med 2005;46:1481-7. 18. Pan T, Mawlawi O, Luo D, Liu HH, Chi P, Mar MV, et al. Attenuation correction of PET cardiac data with low-dose average CT in PET/CT. Med Phys 2006;33:3931-8. 19. Alessio AM, Kohlmyer S, Branch K, Chen G, Caldwell J, Kinahan P. Cine CT for attenuation correction in cardiac PET/CT. J Nucl Med 2007;48:794-801. 20. Mok GSP, Sun T, Huang TC, Vai MI. Interpolated average CT for attenuation correction in PET-A SIMULATION STUDY. IEEE T Bio-Med Eng 2013;60:1927-34. 21. Sun T, Wu TH, Wang SJ, Yang BH, Wu NY, Mok GSP. Low dose interpolated average CT for thoracic PET/CT attenuation correction using an active breathing controller. Med Phys 2013;40: 102507. 22. Huang TC, Mok GSP, Wang SJ, Wu TH, Zhang G. Attenuation correction of PET images with interpolated average CT for thoracic tumors. Phys Med Biol 2011;56:2559-67. 23. Segars WP, Sturgeon G, Mendonca S, Grimes J, Tsui BMW. 4D XCAT phantom for multimodality imaging research. Med Phys 2010;37:4902-15. 24. Yoo TS, Ackerman MJ, Lorensen WE, Schroeder W, Chalana V, Aylward S, et al. Engineering and algorithm design for an image processing API: A technical report on ITK—The insight toolkit. Stud Health Technol 2002;85:586-92. 25. Lujan AE, Balter JM, Ten Haken RK. A method for incorporating organ motion due to breathing into 3D dose calculations in the liver: Sensitivity to variations in motion. Med Phys. 2003;30:2643-9. 26. Thielemans K, Tsoumpas C, Mustafovic S, Beisel T, Aguiar P, Dikaios N, et al. STIR: Software for tomographic image reconstruction release 2. Phys Med Biol. 2012;57:867-83. 27. Wong JW, Sharpe MB, Jaffray DA, Kini VR, Robertson JM, Stromberg JS, et al. The use of active breathing control (ABC) to reduce margin for breathing motion. Int J Radiat Oncol. 1999;44:911-9. 28. Cerqueira MD, Weissman NJ, Dilsizian V, Jacobs AK, Kaul S, Laskey WK, et al. Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart: A statement for healthcare professionals from the Cardiac Imaging Committee of the Council on Clinical Cardiology of the American Heart Association. J Nucl Cardiol. 2002;9:240-5. 29. Pourmoghaddas A, Klein R, deKemp RA, Wells RG. Respiratory phase alignment improves blood-flow quantification in Rb82 PET myocardial perfusion imaging. Med Phys 2013;40:022503. 30. Nam WH, Ahn IJ, Kim KM, Kim BI, Ra JB. Motion-compensated PET image reconstruction with respiratory-matched attenuation correction using two low-dose inhale and exhale CT images. Phys Med Biol. 2013;58:7355-74. 31. Wells RG, Ruddy TD, DeKemp RA, DaSilva JN, Beanlands RS. Single-phase CT aligned to gated PET for respiratory motion correction in cardiac PET/CT. J Nucl Med. 2010;51:1182-90. 32. Cook RAH, Carnes G, Lee TY, Wells RG. Respiration-averaged CT for attenuation correction in canine cardiac PET/CT. J Nucl Med. 2007;48:811-8. 33. Wu TH, Zhang G, Wang SJ, Chen CH, Yang BH, Wu NY, et al. Low-dose interpolated average CT for attenuation correction in cardiac PET/CT. Nucl Instrum Meth A. 2010;619:361-4.
Biomedical Imaging Laboratory, Department of Electrical and Computer Engineering, Faculty of Science and Technology, University of Macau, Macau, SAR, China b Department of Biomedical Imaging and Radiological Sciences, National Yang Ming University, Taipei, Taiwan c Department of Nuclear Medicine, Taipei Veterans General Hospital, Taipei, Taiwan Received Feb 12, 2015; accepted Apr 1, 2015 doi:10.1007/s12350-015-0140-5
Background. Previously, we proposed interpolated averaged CT (IACT) for improved attenuation correction (AC) in thoracic PET/CT. This study aims to evaluate its feasibility and effectiveness on cardiac PET/CT. Methods. We simulated 18F-FDG distribution using the XCAT phantom with normal and abnormal cardiac uptake. Average activity and attenuation maps represented static PET and respiration average CT (ACT), respectively, while the attenuation maps of end-inspiration/expiration represented 2 helical CTs (HCT). IACT was obtained by averaging the 2 extreme phases and the interpolated phases generated between them. Later, we recruited 4 patients who were scanned 1 hr post 315-428 MBq 18F-FDG injection. Simulated and clinical PET sinograms were reconstructed with AC using (1) HCT, (2) IACT, and (3) ACT. Polar plots and the 17-segment plots were analyzed. Two regions-of-interest were drawn on lesion and background area to obtain the intensity ratio (IR). Results. Polar plots of PETIACT-AC were more similar to PETACT-AC in both simulation and clinical data. Artifacts were observed in various segments in PETHCT-AC. IR differences of HCT as compared to the phantom were up to *20%. Conclusions. IACT-AC reduced respiratory artifacts and improved PET/CT matching similarly to ACT-AC. It is a promising low-dose alternate of ACT for cardiac PET/CT. (J Nucl Cardiol 2015) Key Words: PET/CT Æ cardiac imaging Æ attenuation correction Æ respiratory artifacts
INTRODUCTION The power of integrated PET/CT imaging has been well recognized for oncological applications, and its effectiveness in cardiology is a growing interest in both research and clinical practice. Common cardiac PET 82 perfusion agents include 13NH3, H15 Rb while 2 O, and
Reprint requests: Greta S. P. Mok, Ph. D, Biomedical Imaging Laboratory, Department of Electrical and Computer Engineering, Faculty of Science and Technology, University of Macau, Avenida da Universidade, Taipa, Macau, SAR, China; [email protected], [email protected] 1071-3581/$34.00 Copyright 2015 American Society of Nuclear Cardiology.
the most common tracer for viability assessment is 18FFDG.1 Although SPECT is still the primary imaging modality for myocardial perfusion imaging (MPI), PET emerges as an increasingly popular alternate to MPI SPECT, providing superior image quality as well as diagnostic accuracy especially for obese patients, patients undergoing pharmacologic stress, and equivocal cases.2-6 In addition, quantitative analyses such as myocardial blood flow (MBF), coronary flow reserve (CFR), and left ventricular ejection fraction (LVEF) are feasible on PET/CT, with improved risk stratification and patient management.7,8 The use of CT in PET/CT provides intrinsically registered anatomical information for PET functional
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
images and improves the quantitative accuracy with CTbased attenuation correction (AC). For cardiac imaging, added benefits from CT include calcium scoring9 and coronary CT angiogram (CTA)10 from high-end CT scanners. However, due to the potential mis-registration of the sequential PET and CT scans where the acquisition time of PET spans over multiple respiratory cycles, but conventional helical CT (HCT) scan just spans over few seconds, using the mis-registered CT for PET AC will result in respiratory misalignments and artifacts. While signature curvilinear artifacts are often seen in the lung-diaphragm border, extra cardiac misalignments are sometimes observed in the tissues around the lung and left-ventricle interface, where PET myocardial uptake can be overlaid on the lung tissues on the CT images.11 Both effects can potentially affect the further quantitative cardiac analysis. For MPI PET/CT, more than 40% of the studies have artifactual defects in the cardiac region when no interventions are taken to address the PET/CT misalignment, causing significant diagnostic errors.11 Different types of mismatch lead to artifacts and increases in myocardial non-uniformity.12,13 Mis-registration of the stress PET/CT also affect the global and regional myocardial blood flow estimation.14,15 Besides simple manual or automatic registration based on the outline of the heart for the misaligned PET and CT as suggested by Martinez-Mo¨ller et al,14 breathing instruction, CT protocols, and gated 4-dimensional (4D) PET/CT have been investigated mostly to reduce the PET/CT misalignments and artifacts so far. A detailed summary for different motion artifact reduction methods can be found from our previous review article.16 Among them, respiration average CT (ACT) which acquired 4D CT in the thoracic region was developed for AC in PET and showed significantly less misalignments and artifacts as compared with conventional HCT-AC.17,18 Alessio et al further evaluated both average and intensity maximum images of 4D CT and
Journal of Nuclear Cardiology
indicated the later method had better alignments between PET and CT.19 The main problem of ACT is relatively high radiation dose. Previously, we developed an interpolated average CT (IACT) method as a low-dose alternate to ACT for AC to reduce PET/CT misalignments. This method requires the aid of an active breathing controller (ABC) to acquire the end-inspiration and end-expiration phases of the patients. Its effectiveness in image quality and quantitative accuracy improvement for thoracic lesions were demonstrated in our previous simulation20 and clinical studies.21,22 In this study, we aim to evaluate the potential of using IACT for PET AC in cardiology applications. Both simulation and patient data were analyzed to compare the performance of IACT, ACT, and HCT.
MATERIALS AND METHODS Simulation Study We simulated a male patient injected with 18F-FDG distribution using the digital 4D Extended Cardiac Torso phantom (XCAT)23 with cardiac motion and respiratory motion of 2, 3, and 4 cm (Figure 1a). Normal cardiac activity uptake as well as an ischemic heart was modeled, where a longitudinal 2-cm and circumferential 60 lesion with 60% of the normal wall uptake was placed at the lateral and inferior wall of the left ventricle (LV) myocardium individually (Figure 1b, c). The respiratory period of 5.9 seconds was binned into 13 phases, with average activity and attenuation maps to represent static PET and ACT, respectively. The attenuation maps of end-inspiration and end-expiration represented 2 different HCTs (HCT-1 and HCT-8). We used B-spline method to calculate the deformation vectors which include lateral, anterior-posterior, and inferiorsuperior displacement for each voxel between HCT-1 and HCT-8 based on the Insight Segmentation and Registration Toolkit (ITK).24 The forward deformation vector, i.e., from HCT-1 to HCT-8, and backward deformation vector, i.e., from
Figure 1. (A) Average 18F-FDG activity map with cardiac motion and respiratory motion of 2 cm. Short-axis images showing lesion placed at (B) lateral wall and (C) inferior wall.
Journal of Nuclear Cardiology
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
Figure 2. (A) Respiratory phase#1 of the attenuation map represented HCT obtained at end-inspiration (HCT-1); (B) respiratory phase#8 of the attenuation map represented HCT obtained at end-expiration (HCT-8); (C) average attenuation map represented ACT, (D) IACT.
Figure 3. Polar plots generated from the noise-free PET short-axis reconstructed images for a normal heart simulation with respiratory motion amplitude of (A) 2 cm; (B) 3 cm; (C) 4 cm, and for a heart with a lesion placed at (D) inferior wall and (E) lateral wall with respiratory motion of 2 cm. The ROIs drawn for calculating IR were marked in the red squares. HCT-8 to HCT-1, were then used to generate 10 interpolated phases using an empirical diaphragmatic movement function.25 The IACT was obtained by averaging the original extreme
phases, i.e., HCT-1 and HCT-8, and the interpolated phases. More details about the IACT generation can be found in our previous work.21
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
To simulate noisy CT data, CT projections were generated from different AC maps using an analytical projector and then added with Gaussian noise. The projections were then reconstructed with filtered back-projection method to get the reconstructed CT images with similar noise level as in the clinical CT data for further AC in PET (Figure 2), where the noise measurement was based on the standard deviation of a region-of-interest (ROI) drawn in a homogenous liver region on the clinical CT patient data.22 Noise-free and noisy sinograms with attenuation modeling were generated for 2 bed positions to cover the thoracic region and reconstructed with different AC maps (ACT, HCTs, and IACTs) by Software for Tomographic Image Reconstruction (STIR),26 using ordered-subset expectation maximization (OS-EM) method with up to 300 updates. Noisy PET sinogram was generated by adding Poisson noise on the noise-free data based on a count level of *9.5 M, representing a 20 minutes whole-body PET acquisition. The simulation geometry was based on the GE Discovery STE PET scanner.
Patient Study Between October 2012 and December 2012, we recruited 4 male patients with age from 41 to 86 who had their scheduled
Journal of Nuclear Cardiology
whole-body PET/CT scans under the ethics approval of the Institutional Review Board of Taipei Veterans General Hospital. Written informed consents were obtained from all patients. They were scanned 1 hour post 315 to 428 MBq 18F-FDG injection on a clinical PET/CT scanner (Discovery VCT, GE Medical Systems, Milwaukee, WI), which operated in 3dimensional (3D) mode with transaxial field-of-views (FOVs) of 70 and 50 cm for PET and CT, respectively. During the whole-body PET scans, 7 bed positions with 3 minutes/bed were acquired for each patient, and PET sinograms were reconstructed with AC using (1) standard shallow free breathing HCT (120 kV, smart mA (30-150 mA), 0.984:1 pitch); (2) IACT generated from 2 end-inspiration and end-expiration breath-hold HCTs (120 kV, 10 mA) aided by an ABC prototype; and (3) ACT (120 kV, 10 mA, 5.9 seconds duration). The home-made ABC system integrates a spirometer, an air mask, a tube-valve system, and a C??-based acquisition program to ensure the patients could hold their breaths at the desired end-inspiration and end-expiration phases, with breathing amplitude similar to the free breathing state.21,27 The IACT was then generated using the same procedure as mentioned in the simulation with the captured extreme CT phases. All data were reconstructed on the scanner workstation using 3D OSEM method with 2 iterations and 28 subsets. Besides AC, the
Figure 4. (A) The PET short-axis reconstructed images with AC using different CT protocols for patient#1; (B) the corresponding polar plots; (C) polar plots for patient#2.
RESULTS The polar plots of PETIACT-AC were more similar to PETACT-AC from visual assessment in both simulations (Figure 3) and clinical data (Figure 4). Artifacts were observed in PETHCT-AC in various locations including anterior, anterolateral, anteroseptal, and inferior segments of LV. In simulations, the IR differences were close for PETIACT-AC and PETACT-AC, while it can reach *20% for PETHCT-8-AC (Table 1). The seventeen-segment plots of simulations and clinical data indicating the RD as compared to the reference are shown in Figures 5 and 6. The RD was substantially higher with more nonuniformity observed for HCT-AC. For the PET/CT fusion images, significant misalignment was observed at the interface between the lung and left-ventricle in PETHCT-AC/HCT (Figure 7). DISCUSSION The mismatch of emission data and CT-based transmission map was a well-recognized problem, no matter for gated or static emission acquisition. Respiratory gating in PET reduces substantial motion blurring and phase matched CT from 4D or virtual 4D CT was proposed to further improve the PET-CT
0.85 0.88 0.86 0.92 0.98 1.01 0.95 1.08 0.96 0.98 0.93 1.06 3.86 3.87 3.79 3.88 3.47 3.50 3.33 3.59 3.48 3.52 3.26 3.63 3.28 3.41 3.25 3.57 3.42 3.53 3.17 3.87 3.35 3.46 3.03 3.87 *%diff = (IRrecon - IRphantom)/IRphantom 9 100%
The clinical PET reconstructed images with AC using different AC maps, i.e., PETHCT-AC, PETACT-AC, and PETIACT-AC were also registered with associated CT for further visual assessment.
4 cm
ð2Þ
3 cm
ROIlesion : ROIbg
Table 1. Intensity ratios (IRs) for different AC schemes and motion amplitudes
IR ¼
2.83 3.28 -0.59 5.26 7.02 7.04 4.00 9.11 11.98 12.49 11.36 15.48
The original phantom polar plot was used as the reference in simulation, while the polar plot generated from PETACT-AC was used as the reference in patient study respectively. In simulation, two ROIs were drawn in lesion (ROIlesion) and background (ROIbg) area in the polar plots to calculate the intensity ratio (IR, Figure 3d, e):
0.81 0.81 0.78 0.83 0.88 0.88 0.86 0.90 0.78 0.79 0.78 0.81
ð1Þ
3.86 3.87 3.79 3.87 3.47 3.50 3.33 3.59 3.48 3.52 3.26 3.63
I recon Iref 100 %: RD ¼ Iref
3.12 3.14 2.96 3.21 3.07 3.09 2.85 3.23 2.72 2.77 2.53 2.93
Polar plots were generated from the reconstructed shortaxis images for both simulations and patient study. Based on the American Heart Association guidelines, seventeen segments were drawn in the polar plots.28 Relative difference (RD) of the average count value of each segment on the reconstruction images using different AC methods (Irecon) and reference images (Iref) was computed to quantify the recovery of the activity concentration:
ACT IACT HCT-1 HCT-8 ACT IACT HCT-1 HCT-8 ACT IACT HCT-1 HCT-8
Data Analysis
2 cm
PET sinograms were corrected for random, scatter, and isotope decay.
-0.09 3.43 0.88 8.23 5.30 8.16 1.96 15.31 8.22 10.25 4.41 19.56
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
Motion AC amplitude scheme ROIlesion_inferior ROIbg_inferior IRinferior %diffinferior* ROIlesion_lateral ROIbg_lateral IRlateral %difflateral*
Journal of Nuclear Cardiology
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
Journal of Nuclear Cardiology
Figure 5. Seventeen-segment analysis based on the noisy polar plots for a normal heart simulation with motion amplitude of (A) 2 cm; B 3 cm and (C) 4 cm. The RD as compared to the phantom is displayed as the grayscale in each segment, with dark red indicates 40% and light yellow indicates 0%.
alignment.29,30 Different PET respiratory phases can also be transformed and registered to a reference CT phase for further reconstruction.31 However, respiratory gating was still not a clinical routine procedure probably due to implementation complexities and the increased noise level in each respiratory bin. While ACT was proven to be an effective AC method for respiration-ungated cardiac PET,32 similar radiation concern arises as 4D CT. Thus, previously we proposed IACT method as an ACT alternate with lower radiation dose. In the clinical study, the estimated mean effective doses were 0.38, 2.26, and 2.01 mSv for IACT, HCT, and ACT, respectively. The IACT reduced 83% and 81% effective dose as compared to HCT and ACT. The robustness of the ABC device was discussed in our previous work.21 Our recruited patients all showed good compliance to the device, with minimal increase of the radiation dose as the lowest tube current, i.e., 10 mA, was used in the study. The results of IACT substantially approached to those of the ACT as compared to HCT
(Figures 4 and 6), indicating the effectiveness of our algorithm and the ABC. Previously patients’ voluntarily breath-hold was used to capture the 2 extreme phases for generating IACT, without using ABC.33 However, we found that the result was highly unpredictable and misalignments still persisted or even exaggerated for some cases. Thus, the ABC device is an essential component for IACT-AC. In the simulation study, we found that the IR performance of HCT-1, i.e., HCT acquired at endinspiration phase, was superior to HCT-8, i.e., HCT acquired at end-expiration phase and even ACT. That might due to the fact that the diaphragm was at a lower position for end-inspiration state and would cause less attenuation to the heart. Less blurring from the HCT might also improve the lesion delectability. However, from the global assessment as in the 17-segment plot, HCT-1 was superior to HCT-8 but was still inferior to both ACT and IACT. The general matching of static PET and CT was markedly improved for the average CT techniques (Figure 7).
Journal of Nuclear Cardiology
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
Figure 6. Seventeen-segment analysis based on the clinical data as shown in Figure 4, with (A) & (B) generated from patient#1 and (C) & (D) generated from patient#2. The RD as compared to the PETACT-AC is displayed as the grayscale in each segment, with dark red indicates 40% and light yellow indicates 0%.
Figure 7. Fusion images of (A) PETACT-AC /ACT; (B) PETIACT-AC /IACT; and (C) PETHCT-AC /HCT for patient#1.
In this study, only static 18F-FDG PET scan was evaluated. The ECG gating 18F-FDG scan is also a common clinical procedure, using a single breath-hold or shallow free breathing HCT as the attenuation map for different PET phases. Further evaluation in using ACT/IACT for cardiac gated PET scan is warranted. Interpolated CT (ICT), a precursor of IACT, might be more suitable for AC in respiratory gated PET scan and is currently under investigation by our group. More evaluations on using IACT-AC on myocardial perfusion PET are also warranted.
CONCLUSION Our simulation and clinical studies accordantly showed that the respiratory artifacts can be reduced, while the uniformity and the PET/CT matching can be improved with ACT/IACT-AC in cardiac PET, as compared to standard HCT-AC. The IACT is a good low-dose alternate of ACT and has the potential to improve cardiac PET/CT imaging. Acknowledgments
NEW KNOWLEDGE GAINED
• HCT-AC leads to artifacts in anterior, anterolateral, anteroseptal, and inferior segments of LV, degraded cardiac uniformity, and lesion quantitation on PET cardiac images. • Respiratory average CT, i.e., ACT and IACT, can reduce the PET/CT image mismatch, reduce artifacts, and improve quantitation for cardiac PET, according to the polar plots and 17-segment analysis. • Performances of IACT and ACT are similar for cardiac PET, while IACT shows lower estimated dose.
The authors would like to thank Mr. Tao Sun from Nuclear Medicine and Medical Imaging Research Center at K.U. Leuven and Ms Nien-Yun Wu from the Department of Nuclear Medicine at Taipei Veterans General Hospital for their technical and experimental assistance. This work was supported by research grants (MYRG185(Y4-L3)-FST11-MSP and MYRG077(Y3-L2)-FST12-MSP) from University of Macau, Macau, and a research grant (MOST 103-2314-B195-001-MY3) from the Ministry of Science and Technology of Taiwan, Taiwan.
Disclosure None.
Mok et al Interpolated average CT for cardiac PET/CT attenuation correction
References 1. Matsunari I, Taki J, Nakajima K, Tonami N, Hisada K. Myocardial viability assessment using nuclear imaging. Ann Nucl Med 2003;17:169-79. 2. Heller GV, Calnon D, Dorbala S. Recent advances in cardiac PET and PET/CT myocardial perfusion imaging. J Nucl Cardiol 2009;16:962-9. 3. Di Carli MF, Dorbala S, Meserve J, El Fakhri G, Sitek A, Moore SC. Clinical myocardial perfusion PET/CT. J Nucl Med 2007;48:783-93. 4. Knaapen P, de Haan S, Hoekstra OS, Halbmeijer R, Appelman YE, Groothuis JGJ, et al. Cardiac PET-CT: advanced hybrid imaging for the detection of coronary artery disease. Neth Heart J 2010;18:90-8. 5. Bateman TM, McGhie AI, O’Keefe JH, Thompson RC, Moutray KL, Bryngelson JR, et al. High clinical value of follow-up myocardial perfusion PET in patients with a diagnostically indeterminate myocardial perfusion SPECT study. Circulation 2003;108:454. 6. Bateman TM, Heller GV, McGhie AI, Friedman JD, Case JA, Bryngelson JR, 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;13:24-33. 7. Shah AR, Lertsburapa K, Ahlberg A, Katten D, Volker L, Cullom SJ, et al. What is the prognostic value of myocardial perfusion imaging using rubidium-82 positron emission tomography in obese patients with known or suspected CAD? J Am Coll Cardiol 2008;51:A173-A. 8. Yoshinaga K, Chow BJW, Williams K, Chen L, Dekemp RA, Garrard L, et al. What is the prognostic value of myocardial perfusion imaging using rubidium-82 positron emission tomography? J Am Coll Cardiol 2006;48:1029-39. 9. Burkhard N, Herzog BA, Husmann L, Pazhenkottil AP, Burger IA, Buechel RR, et al. Coronary calcium score scans for attenuation correction of quantitative PET/CT N-13-ammonia myocardial perfusion imaging. Eur J Nucl Med Mol I 2010;37:517-21. 10. Hacker M, Jakobs T, Matthiesen F, Vollmar C, Nikolaou K, Becker C, et al. Comparison of spiral multidetector CT angiography and myocardial perfusion imaging in the noninvasive detection of functionally relevant coronary artery lesions: First clinical experiences. J Nucl Med 2005;46:1294-300. 11. Gould KL, Pan T, Loghin C, Johnson NP, Guha A, Sdringola S. Frequent diagnostic errors in cardiac PET/CT due to misregistration of CT attenuation and emission PET images: A definitive analysis of causes, consequences, and corrections. J Nucl Med 2007;48:1112-21. 12. McQuaid S, Hutton B. Sources of attenuation-correction artefacts in cardiac PET/CT and SPECT/CT. Eur J Nucl Med Mol Imaging 2008;35:1117-23. 13. 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;45:1029-39. 14. Martinez-Mo¨ller A, Souvatzoglou M, Navab N, Schwaiger M, Nekolla SG. Artifacts from Misaligned CT in Cardiac Perfusion PET/CT Studies: Frequency, Effects, and Potential Solutions. J Nucl Med 2007;48:188-93. 15. Rajaram M, Tahari AK, Lee AH, Lodge MA, Tsui B, Nekolla S, et al. Cardiac PET/CT misregistration causes significant changes in estimated myocardial blood flow. J Nucl Med 2013;54:50-4.
Journal of Nuclear Cardiology
16. Sun T, Mok GS. Techniques for respiration-induced artifacts reductions in thoracic PET/CT. Quant Imaging Med Surg 2012;2:4652. 17. Pan T, Mawlawi O, Nehmeh SA, Erdi YE, Luo D, Liu HH, et al. Attenuation Correction of PET Images with Respiration-Averaged CT Images in PET/CT. J Nucl Med 2005;46:1481-7. 18. Pan T, Mawlawi O, Luo D, Liu HH, Chi P, Mar MV, et al. Attenuation correction of PET cardiac data with low-dose average CT in PET/CT. Med Phys 2006;33:3931-8. 19. Alessio AM, Kohlmyer S, Branch K, Chen G, Caldwell J, Kinahan P. Cine CT for attenuation correction in cardiac PET/CT. J Nucl Med 2007;48:794-801. 20. Mok GSP, Sun T, Huang TC, Vai MI. Interpolated average CT for attenuation correction in PET-A SIMULATION STUDY. IEEE T Bio-Med Eng 2013;60:1927-34. 21. Sun T, Wu TH, Wang SJ, Yang BH, Wu NY, Mok GSP. Low dose interpolated average CT for thoracic PET/CT attenuation correction using an active breathing controller. Med Phys 2013;40: 102507. 22. Huang TC, Mok GSP, Wang SJ, Wu TH, Zhang G. Attenuation correction of PET images with interpolated average CT for thoracic tumors. Phys Med Biol 2011;56:2559-67. 23. Segars WP, Sturgeon G, Mendonca S, Grimes J, Tsui BMW. 4D XCAT phantom for multimodality imaging research. Med Phys 2010;37:4902-15. 24. Yoo TS, Ackerman MJ, Lorensen WE, Schroeder W, Chalana V, Aylward S, et al. Engineering and algorithm design for an image processing API: A technical report on ITK—The insight toolkit. Stud Health Technol 2002;85:586-92. 25. Lujan AE, Balter JM, Ten Haken RK. A method for incorporating organ motion due to breathing into 3D dose calculations in the liver: Sensitivity to variations in motion. Med Phys. 2003;30:2643-9. 26. Thielemans K, Tsoumpas C, Mustafovic S, Beisel T, Aguiar P, Dikaios N, et al. STIR: Software for tomographic image reconstruction release 2. Phys Med Biol. 2012;57:867-83. 27. Wong JW, Sharpe MB, Jaffray DA, Kini VR, Robertson JM, Stromberg JS, et al. The use of active breathing control (ABC) to reduce margin for breathing motion. Int J Radiat Oncol. 1999;44:911-9. 28. Cerqueira MD, Weissman NJ, Dilsizian V, Jacobs AK, Kaul S, Laskey WK, et al. Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart: A statement for healthcare professionals from the Cardiac Imaging Committee of the Council on Clinical Cardiology of the American Heart Association. J Nucl Cardiol. 2002;9:240-5. 29. Pourmoghaddas A, Klein R, deKemp RA, Wells RG. Respiratory phase alignment improves blood-flow quantification in Rb82 PET myocardial perfusion imaging. Med Phys 2013;40:022503. 30. Nam WH, Ahn IJ, Kim KM, Kim BI, Ra JB. Motion-compensated PET image reconstruction with respiratory-matched attenuation correction using two low-dose inhale and exhale CT images. Phys Med Biol. 2013;58:7355-74. 31. Wells RG, Ruddy TD, DeKemp RA, DaSilva JN, Beanlands RS. Single-phase CT aligned to gated PET for respiratory motion correction in cardiac PET/CT. J Nucl Med. 2010;51:1182-90. 32. Cook RAH, Carnes G, Lee TY, Wells RG. Respiration-averaged CT for attenuation correction in canine cardiac PET/CT. J Nucl Med. 2007;48:811-8. 33. Wu TH, Zhang G, Wang SJ, Chen CH, Yang BH, Wu NY, et al. Low-dose interpolated average CT for attenuation correction in cardiac PET/CT. Nucl Instrum Meth A. 2010;619:361-4.
