RT Journal Article SR Electronic T1 Correction of respiration induced motion artifacts and activity-attenuation mismatch in free-breathing PET/CT JF Journal of Nuclear Medicine JO J Nucl Med FD Society of Nuclear Medicine SP 1346 OP 1346 VO 58 IS supplement 1 A1 Jizhe Wang A1 Tao Feng A1 Yun Dong A1 Hongdi Li YR 2017 UL http://jnm.snmjournals.org/content/58/supplement_1/1346.abstract AB 1346Objectives: In PET/CT imaging, respiratory motion of the patient introduces motion artifacts in CT images acquired by 16-ring helical CT scanner. Due to short axial scout, the free-breathing CT acquisition provides a mixed-phase image with different trans-axial slices corresponding to different respiratory gates. Moreover, activity-attenuation mismatch is observed in reconstructed PET images of all respiratory gates. The aim of this study is to correct for both undesirable effects through B-spline image registration of respiratory gated PET data and mutual information based PET-CT mismatch correction.Methods: Clinical data were acquired by a commercial PET/CT scanner featuring a 16-ring CT system and a high-resolution PET system. The PET data was divided into five respiratory frames according to respiratory motion amplitude using a previously developed data-driven respiratory gating method. The attenuation map from CT scan was heavily burred in axial direction before use in attenuation correction of the gated PET data to reduce activity-attenuation mismatch. The reconstructed PET images with no obvious mismatch effects were used for motion vector fields (MVF) estimation. MVFs from every other frame to frame #5 were estimated using a B-spline based multi-resolution image registration algorithm whose cost function consists of summed square error between two images and smoothness constraint of the MVF. For each transversal slice, the mutual information between the original attenuation map and PET image of each gate was calculated and used as a similarity score to determine the respiratory phase of the attenuation map in that slice. High mutual information indicates good match between the attenuation map and the activity distribution. Among all the respiratory frames of PET images, the frame that achieved the highest mutual information with attenuation map was identified as the corresponding respiratory frame of the attenuation map for the specific transversal slice. As a result, five gates of attenuation map, each of which contains incomplete information of a certain respiratory phase, were de-interlaced from the original mixed-phase attenuation map. The five frames of incomplete attenuation maps were transformed to frame #5 using the MVFs estimated from gated PET images with necessary interpolation to obtain a complete attenuation map at frame #5. Attenuation map for other gates were transformed from the attenuation map of frame #5 with reversed MVFs. Respiratory gated attenuation maps corrected with proposed methods showed reduced motion artifacts. PET images of different gates overlapped with attenuation map of corresponding gate demonstrated minimized mismatch compared to those with the original attenuation map.Results: Respiratory gated attenuation maps corrected with proposed methods showed reduced motion artifacts. PET images of different gates overlapped with attenuation map of corresponding gate demonstrated minimized mismatch compared to those with the original attenuation map.Conclusion: In this study, we proposed to use mutual information as similarity measurement to identify different respiratory phases in free-breathing CT image. Complete attenuation map of each respiratory gate was created by transforming incomplete attenuation map with MVFs estimated from gated PET images by minimizing mean squared error and smoothness constraint. Reduction of motion artifacts and activity-attenuation mismatch was observed from clinical data processed with the proposed method. Research Support: