Advances in 4D Gated Cardiac PET Imaging for Image Quality Improvement and Cardiac Motion and Contractility Estimation
Quantitative four-dimensional (4D) image reconstruction methods with respiratory and cardiac motion compensation are an active area of research in ECT imaging, including SPECT and PET. They are the extensions of three-dimensional (3D) statistical image reconstruction methods with iterative algorithms that incorporate accurate models of the imaging process and provide significant improvement in the quality and quantitative accuracy of the reconstructed images as compared to that obtained from conventional analytical image reconstruction methods. The new 4D image reconstruction methods incorporate additional models of the respiratory and cardiac motion of the patient to reduce image blurring due to respiratory motion and image noise of the cardiac-gated frames of the 4D cardiac-gated images. We describe respiratory motion estimation and gating method based on patient PET list-mode data. The estimated respiratory motion is applied to the respiratory gated data to reduce respiratory motion blur. The gated cardiac images derived from the list-model data are used to estimate cardiac motion. They are then used in the cardiac-gated images summing the motion-transformed cardiac-gated images for significant reduction in the gated images noise. Dual respiratory and cardiac motion compensation is achieved by combining the respiratory and cardiac motion compensation steps. The results are further significant improvements of the 4D gated cardiac PET images. The much improved gated cardiac PET image quality increases the visibility of anatomical details of the heart, which can be explored to provide more accurate estimation of the cardiac motion vector field and cardiac contractility.
Keywords4D gated cardiac PET 4D image reconstruction methods Respiratory and cardiac motion estimation and compensation
The development of quantitative image reconstruction in medical imaging, including emission computed tomography (ECT) and x-ray CT [1, 2], has recently shifted from three-dimensional (3D) to four-dimensional (4D), i.e., the inclusion of the time dimension. There are two major goals for this development. First is to reduce reconstructed image artifacts due to patient motion. In particular, compensation of involuntary patient motion, e.g., respiratory motion, that causes resolution loss has received much attention [3, 4, 5, 6]. Second is to improve the temporal resolution of dynamic images for improved detection of global and regional motion abnormalities [7, 8]. An important application is gated myocardial perfusion (MP) ECT imaging. Despite extensive research in other imaging modalities over the last two decades, MP ECT, especially gated SPECT and more recently PET, has continued to be the major biomedical imaging technique for the assessment of MP in clinical practice. The potential of extracting additional quantitative information, such as abnormalities from existing data without additional clinical studies, radiation dose or discomfort to the patients, has great significance in biomedical imaging [9, 10, 11, 12, 13].
The long-term goal of the study is to integrate the two aforementioned goals of the current quantitative 4D imaging reconstruction methods, i.e., to improve the quality and quantitative accuracy of the 4D cardiac gated MP PET images while reducing the blurring caused by respiratory motion (RM) and cardiac motion (CM). This is in addition to compensation of other image degrading factors, e.g., statistical noise, photon attenuation and scatter, and collimator-detector blur, to improve both spatial and temporal resolution. In this work, we present the development of a data-driven RM estimation method and quantitative 4D statistical image reconstruction methods that compensate for RM and CM separately, and for dual respiratory and cardiac (R&C) motion for improved lung and cardiac PET imaging. We hypothesize that by applying a statistical 4D image reconstruction method that accurately compensates for RM and CM and other image degrading factors, we would be able to minimize image artifacts caused by the image degrading factors, improve image resolution and reduce image noise. This would result in two significant clinical benefits, i.e., (a) reduction of false positives and false negatives for improved diagnosis, and (b) reduction of imaging time and/or radiation dose to the patient.
In addition, the much improved 4D gated cardiac PET image quality increases the visibility of details of cardiac structures. The information can be explored in a feature-based motion estimation method to determine the cardiac motion vector field and cardiac contractility.
1.2 Materials and Methods
1.2.1 Data-Driven Respiratory Motion Detection and Gating Method
From the estimated RM signal in Fig. 1.1c, we divided the list-mode data into six equal-count respiratory frames, each of which is further divided into eight cardiac-gated frames using the ECG R-wave markers. The result was a full set of dual six-frame respiratory gated and eight-frame cardiac-gated dataset. We then applied the RM compensation method described in Sect. 1.2 to the six-frame respiratory-gated dataset.
1.2.2 4D PET Image Reconstruction Methods with Attenuation, and Respiratory and Cardiac Motion Compensation
The 4D PET image reconstruction methods used in this study were applied to the respiratory-gated and cardiac-gated projection data. Specifically, for the 4D PET image reconstruction method with dual R&C motion compensation, we divided the acquired list-mode data into six equal-count respiratory frames each with eight cardiac-gated frames as described in Sect. 1.2.1. Image reconstructions without attenuation correction were performed on this dataset to estimate the RM in lung PET studies and both RM and CM in cardiac PET studies. A special feature of our method was the modeling of the RM-induced deformations of the PET image and CT-based attenuation map in RM estimation and during PET image reconstruction for accurate and artifact-free attenuation corrected PET images.
184.108.40.206 4D PET Image Reconstruction with Respiratory Motion and Attenuation Compensation
In a practical implementation of the above method , the RM was estimated from the 4D respiratory- gated PET images obtained without attenuation correction. The estimated RM was used in the 4D image reconstruction shown in Fig. 1.2 without further update. It provided respiratory-gated attenuation effect that matches the respiratory-gated PET images for accurate attenuation compensation.
220.127.116.11 4D Image Reconstruction with Cardiac Motion Compensation
18.104.22.168 4D Image Reconstruction with Dual Respiratory and Cardiac Motion Compensation
The 4D image reconstruction with dual R&C motion compensation was achieved by combining the RM and CM compensation described in Sects. 22.214.171.124 and 126.96.36.199. After estimating the accurate RM and respiratory gated attenuation maps based on Sect. 188.8.131.52, 48-frame dual R&C gated images were obtained. For each cardiac gate, the RM compensation described in Sect. 184.108.40.206 was used. The result was RM compensated cardiac-gated images. Cardiac motion compensation was achieved by applying the same approach in Sect. 220.127.116.11. The resultant eight-frame gated cardiac images shown in Fig. 1.3 thus included both RM and CM compensation.
1.2.3 Evaluation of the 4D PET Image Reconstruction with Respiratory and Attenuation Compensation
We evaluated the 4D image reconstruction method with respiratory and attenuation compensation to two clinical applications. They were the detection of small lung lesions and the improvement of image quality in gated cardiac PET images. In the lung lesion detection study, we used realistic simulated 4D respiratory gated lung PET projection data. In the gated cardiac study, patient data from a 13NH3 MP PET study and a 18F-FDG cardiac PET study were used. The goal was to assess the reduction of image resolution from blurring due to RM.
18.104.22.168 Evaluation Using Realistic Simulated PET Study with Small Lung Lesions
We evaluated the 4D PET image reconstruction with respiratory and attenuation compensation method using a realistic Monte Carlo (MC) simulated PET dataset from the 4D XCAT (eXtended CArdiac Torso) phantom . The 4D XCAT phantom is an extension of the 4D NCAT (Nurbs-based CArdiac Torso) phantom , which provides realistic models of the anatomical structures of the entire human body based on the visible human data . In addition, the 4D XCAT phantom includes realistic models of normal RM based on respiratory-gated CT data , and normal cardiac motion based on tagged MRI data. The cardiac motion model in the new 4D XCAT is based on state-of-the-art high-resolution cardiac-gated CT and tagged MRI data . A 4D activity distribution phantom that modeled the uptake of the PET tracer in the different organs and a corresponding 4D attenuation coefficient distribution phantom that modeled the attenuation of different organs at the 511 keV photon energy were generated based on the 4D XCAT phantom. In addition, three small lung lesions with increased activity uptakes were inserted at different locations in the lung. The 4D activity distribution also served as the truth in the quantitative evaluation study.
Realistic respiratory-gated PET projection data were generated from the 4D activity and attenuation distributions using a combined SimSET  and GATE  MC simulation software that took advantage of the high efficiency of the former in computing the photon transport in the voxelized phantom and the ability of the latter to model the complex detector geometry and imaging characteristics of a clinical GE PET system . The 4D PET image reconstruction method with RM and attenuation compensation was applied to the simulated RM-gated projection data. The results were compared to those obtained with conventional 3D and 4D image reconstruction methods without motion compensation.
22.214.171.124 Evaluation Using Data from Clinical Gated Cardiac PET Studies
We also evaluated the clinical efficacy of the 4D image reconstruction method with RM and attenuation compensation using clinical 13NH3 MP PET and 18F-FDG cardiac PET data. A GE Discovery VCT (RX) PET/CT system was used in the patient studies. Prior to the PET scan, a low-dose CT scan was acquired from the patient. In the 13NH3 MP PET study, ~370 MBq of 13NH3 was infused intravenously as a bolus over 10 s. List-mode PET data were acquired for 20 min. In the 18F-FDG cardiac PET study of a different patient, ~370 MBq of 18F-FDG was administered through IV injection. A list-mode PET data acquisition was performed ~60 min post injection. The 4D image reconstruction method with RM and attenuation compensation as described in Sect. 1.2.1 were applied to the acquired list-mode data. The resultant MP PET and cardiac PET images were compared to those obtained with the conventional image reconstruction method without RM compensation. Specifically, they were evaluated for improved lung lesion detection from the reduction of resolution loss due to RM blur.
1.2.4 Evaluation of the 4D PET Image Reconstruction Method with Dual Respiratory and Cardiac Motion Compensation
The evaluation of the 4D PET image reconstruction method with dual R&C motion compensation was performed on the same clinical 13NH3 MP PET and 18F-FDG cardiac PET datasets used in Sect. 126.96.36.199. Here, the goal was to assess the improvement of the quality of the gated cardiac PET images in terms of image resolution and image noise.
1.3 Results and Discussion
1.3.1 Improvement of Small Lung Lesion Detection with Respiratory and Attenuation Compensation
1.3.2 Improvement of Gated Cardiac PET Images with Respiratory Motion and Attenuation Compensation
1.3.3 Improvement of Gated Cardiac PET Images with Dual Respiratory and Cardiac Motion Compensation
Three-dimensional (3D) statistical image reconstruction methods using iterative algorithms and with models of the imaging physics and imaging system characteristics have shown to provide significant improvements in both the quality and quantitative accuracy of static SPECT and PET images. They have led to improved clinical diagnosis and, by trading off the improved image quality, for reduced patient dose and imaging time. In this work, we described newly developed 4D statistical image reconstruction methods that provided RM and CM compensation for further improvement in image quality and quantitative accuracy in PET images. We evaluated the effectiveness of the 4D image reconstruction methods using simulation and patient data.
Our results showed that a 4D image reconstruction method with RM and attenuation compensation provided quantitative lung PET images with reduced resolution loss due to RM blur and improved the detection of small lung lesions. We also evaluated a 4D image reconstruction method with dual R&C motion compensation using data from a clinical 13NH3 MP PET and a clinical 18F-FDG cardiac PET study. The results showed 4D gated cardiac PET images with improved image resolution from RM compensation and much lower image noise level from the CM compensation.
The improved 4D gated cardiac PET images reveal anatomical details, such as the papillary muscle and interventricular sulcus, of the heart that were not possible with conventional 3D image reconstruction methods. The anatomical details allowed the development of feature-based myocardial motion vector estimation methods [33, 34] that overcame the aperture problem in traditional motion estimation methods. The accuracy of CM estimation will be further improved with continued improvement of the 4D image reconstruction methods and of the imaging characteristics in the next generation PET scanners that are coming into the market. It will allow extraction of new information about the contractility of the heart and provide additional diagnostic information for improved patient care.
- 4.Chung A, Camici P, Yang G-Z, editors. List-mode affine rebinning for respiratory motion correction in PET cardiac imaging. Medical imaging and augmented reality. Berlin/Heidelberg: Springer; 2006.Google Scholar
- 5.Chen S, Tsui BMW. Evaluation of a new 4D PET image reconstruction method with respiratory motion compensation in a CHO study. J Nucl Med 2011: 150.Google Scholar
- 6.Chen S, Tsui BMW. Evaluation of a 4D PET image reconstruction method with respiratory motion compensation in a patient study. Society of nuclear medicine annual meeting. San Antonio; 2011: J Nucl Med. 2011. p. 2023.Google Scholar
- 8.Lee T-S, Tsui BMW. Optimization of a 4D space-time gibbs prior in a 4D MAP-RBI-EMReconstruction method for application to gated myocardial perfusion SPECT. Proceeding of the fully three-dimensional image reconstruction meeting in radiology and nuclear medicine. 2009. p. 122.Google Scholar
- 13.Lee T-S, Tsui BMW. Evaluation of corrective reconstruction method for reduced acquisition time and various anatomies of perfusion defect using channelized hotelling observer for myocardial perfusion SPECT. IEEE nuclear science symposium and medical imaging conference record. 2010. p. 3523–6.Google Scholar
- 18.Klein GL, Reutter BW, Huesman RH. Data-driven respiratory gating in list mode cardiac PET. J Nucl Med. 1999;40:113p.Google Scholar
- 19.Lamare F, Ledesma Carbayo MJ, Cresson T, Kontaxakis G, Santos A, Cheze Le Rest C, Reader AJ, Visvikis D. List-mode-based reconstruction for respiratory motion correction in PET using non-rigid body transformations. Phys Med Biol. 2007;52:.Google Scholar
- 20.Chen ST, Tsui BMW. Accuracy analysis of image registration based respiratory motion compensation in respiratory-gated FDG oncologcial PET reconstruction. In: IEEE nuclear science symposium & medical imaging conference. Dresden; 2008. p. M06-417.Google Scholar
- 21.Chen S, Tsui BMW. Four-dmiensional OS-EM PET image reconstruction method with motion compensation. In: Fully three-dimensional image reconstruction in radiology and nuclear medicine. Beijing; 2009. p. 373–6.Google Scholar
- 24.Chen S, Tsui BMW. Joint estimation of respiratory motion and PET image in 4D PET reconstruction with modeling attenuation map deformation induced by respiratory motion. J Nucl Med. 2010;51(supplement 2):523.Google Scholar
- 26.Segars WP. Development of a new dynamic NURBS-based cardiac-torso (NCAT) phantom, PhD dissertation, The University of North Carolina, May 2001.Google Scholar
- 28.Segars WP, Mori S, Chen G, Tsui BMW. Modeling respiratory motion variations in the 4D NCAT Phantom. In: IEEE medical imaging conference. 2007. p. M26-356.Google Scholar
- 30.Lewellen TK, Harrison RL, Vannoy S. The simset program. In: Monte Carlo calculations in nuclear medicine, Medical science series. Bristol: Institute of Physics Publication; 1998. p. 77–92.Google Scholar
- 32.Shilov M, Frey EC, Segars WP, Xu J, Tsui BMW. Improved Monte-Carlo simulations for dynamic PET. J Nucl Med Suppl. 2006;47:197.Google Scholar
- 34.Wang J, Fung GSK, Feng T, Tsui BMW. An interventricular sulcus guided cardiac motion estimation method. Conference record of the 2013 I.E. nuclear science symposium and medical imaging conference, Seoul, 2013;October 27–November 2. p. 978–84.Google Scholar
Open Access This chapter is distributed under the terms of the Creative Commons Attribution-Noncommercial 2.5 License (http://creativecommons.org/licenses/by-nc/2.5/) which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.
The images or other third party material in this chapter are included in the work’s Creative Commons license, unless indicated otherwise in the credit line; if such material is not included in the work’s Creative Commons license and the respective action is not permitted by statutory regulation, users will need to obtain permission from the license holder to duplicate, adapt or reproduce the material.