Precision PET attenuation correction for combined PET/MR imaging of cardiac sarcoidosis - PROJECT SUMMARY Combined PET/MR, enabling simultaneous PET and MRI, has the potential to at least match the diagnostic performance of reviewing separate PET/CT and MRI scans, in a single integrated exam of the heart. This will substantially reduce the patient burden, exam costs, and radiation compared to serial imaging on PET/CT and MRI, and benefit the workup of multiple heart diseases. Further, by enabling accurate spatial overlap of PET and MR images, and the ability to generate gold standard metabolic measures from extended dynamic PET scans, PET/MR could outperform interpretation of serial scans. However, quantitative accuracy of tracer uptake on PET images from PET/MR studies frequently is lower than that from reference standard PET/CT. The primary cause of this discrepancy are errors in MR-based corrections for PET signal attenuation. MR- based attenuation corrections suffer from artifacts due to metallic cardiac implants, breathing motion, and limited lung signal, which can all propagate into the reconstructed PET image. This is an important problem, since PET/MR may have limited efficacy in the workup of patients with heart diseases if PET image quality and quantification does not match PET/CT systems. The overall objective of this proposal is to develop a comprehensive PET attenuation correction scheme, using the signal from an external PET source and the patient, to enable precision PET imaging during cardiac PET/MR exams. The method utilizes hardware, image reconstruction, and deep learning algorithms to directly measure and correct for PET attenuation based on physics principles alone, without using prior data or assumptions. The specific aims include: 1) fabricating a device that can rapidly position and remove an external PET source in a PET/MR that is practical and safe for patient imaging, 2) developing algorithms for real-time hardware optimization and deep learning enhancement that ensure robust PET quantification for all exams, and 3) evaluating the proposed attenuation correction scheme on the workup of patients with the infiltrating disease, cardiac sarcoidosis. We focus on cardiac sarcoidosis, as these patients often have a cardiac defibrillator implanted before imaging, particularly challenging attenuation correction methods, and PET and MRI are the standard of care for workup. We will assess our algorithms with phantoms, simulations, and prospective FDG-PET/MR and PET/CT studies of sarcoidosis patients. Successful completion of these aims will enable realization of PET/MR as a comprehensive integrated workup of cardiac sarcoidosis and heart diseases overall.
