TY - JOUR T1 - Non-transmural myocardial perfusion defect detection using 5D respiratory and cardiac motion-corrected PET imaging JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 488 LP - 488 VL - 54 IS - supplement 2 AU - Jing Tang AU - Arman Rahmim Y1 - 2013/05/01 UR - http://jnm.snmjournals.org/content/54/supplement_2/488.abstract N2 - 488 Objectives Respiratory (R) and cardiac (C) motion degrades cardiac PET imaging, affecting the detectability of myocardial perfusion (MP) defects. The goal is to study the effect of joint R and C motion correction on non-transmural MP defect detection. Methods We developed a 5D joint R & C motion-corrected image reconstruction method. For each C gate, integrated 4D R motion-corrected image reconstruction was performed to the R-gated data, with the end-expiratory frame as the reference. Using the C motion vector fields estimated from the R motion-corrected C gates, we warped and summed the C-gated images, with the end-diastolic frame as the reference. Using the XCAT phantom, we simulated three dual-gated Rb-82 PET datasets, the 1st with normal myocardium, the 2nd with a transmural MP defect, and the 3rd with a non-transmural MP defect. Each dataset had 100 noise realizations of 5 R and 8 C gated frames, with the organ activity and the noise level representing a typical patient Rb-82 scan. The transmural and the non-transmural defects were simulated on the same region of the left ventricle and they carried the same level of perfusion reduction. Using a channelized Hotelling observer, we performed receiver operating characteristic (ROC) analysis for MP defect detection on reconstructed images with and without R & C motion correction. Results From MP images reconstructed from all the R and C frame data without motion correction, the ROC analysis results in an area under the curve (AUC) value of .85 ± .04 for the transmural defect and an AUC value of .68 ± .05 for the non-transmural defect. The AUC values estimated from images obtained from the proposed R & C motion compensated all frame data are .96 ± .02 for the transmural defect and .85 ± .04 for the non-transmural defect. Conclusions The proposed R & C motion-corrected image reconstruction method improves MP defect detection for both transmural and non-transmual defects significantly. The approach may have promising applications for detecting especially non-transmural MP defects. ER -