Research Spotlight Archive
Title: Joint Cardiac and Respiratory Motion Correction and Super-Resolution in Coronary PET/CT
Participants: Boston University – Sonal Ambwani (PhD ’11) and Professor W. Clem Karl
Massachusetts General Hospital – Dr. Ahmet Tawakol and Dr. Homer Pien
Funding: Bernard M. Gordon Center for Subsurface Sensing & Imaging Systems (CenSSIS)
Background: Coronary artery diseases (CAD) such as Atherosclerosis is the leading cause of mortality in industrialized nations. It is characterized by chronic inflammation and plaque formation in the arterial walls due to the deposition of activated macrophage white blood cells, low-density lipoproteins, and calcium. If not properly diagnosed and treated, this plaque can lead to severe complications resulting in myocardial infarction. Cardiac catheterization is the current gold standard for assessing the severity of coronary artery stenosis, but this technique provides no information about the physiological state of the coronary walls. Non-invasive techniques such as cardiac computed tomography (CT) provide information on both stenosis and plaque composition but yield no insight on the degree of inflammation. Positron emission tomography (PET) has been shown to have the potential of imaging inflammation in large and static blood vessels such as the carotid artery in the neck. However, PET images of the heart have severe blurring induced by the presence of cardiac and respiratory motion during the 15-20 minutes of PET acquisition. Figure 1 illustrates the PET/CT cardiac imaging problem.
Description: The overall aim of this research project is to allow direct, non-invasive imaging of developing coronary plaques. This goal can be accomplished by enabling high resolution, high signal-to-noise (SNR) imaging of coronary artery inflammation through PET-CT. This is achieved by jointly compensating for both cardiac and respiratory motion and performing super-resolution reconstruction of the PET. X-ray CT images are taken in a breath-hold state. Thus, they are used to provide respiration-free cardiac-motion information. It is safe to assume that the PET and CT volumes are in perfect alignment. This cardiac motion information is then incorporated in a unified PET reconstruction functional which jointly estimates the residual respiratory motion, corrects for both phase-aligned cardiac and respiratory motion and provides a super-resolved estimate of the PET reconstruction. A key feature of this approach is that we make use of all available data which leads to the preservation of SNR. We have termed our technique as Data-Domain Cardiac Shape Tracking and Adjustment for Respiration or D-CSTAR.
Results: The experiments have been performed on the Extended Cardiac Torso (XCAT) generated cardiac CT and PET activity phantoms. In Figure 2 and 3 we present the comparison of the D-CSTAR approach with the conventional PET reconstruction and two of the methods proposed previously in literature. The proposed D-CSTAR method has proved to outperform these methods both qualitatively and quantitatively in reconstructing a well-defined, localized lesion.
Publications: S. Ambwani, S. Cho, W. C. Karl, A. Tawakol, and H. Pien, “A feasibility study of joint respiratory and cardiac motion correction for coronary PET/CT imaging,” Biomedical Imaging: From Nano to Macro, IEEE International Symposium on Biomedical Imaging, pp. 935-938, June 2009.
S. Ambwani, W. C. Karl, A. Tawakol, and H. Pien, “Joint cardiac and respiratory motion correction and super-resolution in coronary PET/CT,” submitted to 2011 International Symposium on Biomedical Imaging.