Abstract
1633
Objectives: A solution to the problem of respiratory motion in cardiac PET/CT studies is proposed.
Methods: Respiratory gating is used to get 8 separate PET respiratory phases. A discontinuity preserving multi-scale optical flow method is then applied to get motion vectors between all phases. Also, the single CT-based attenuation map is deformed with the motion vectors to get 8 attenuation maps. Lastly, a complete reconstruction step is used which considers the PET motion vectors and the attenuation-maps during the reconstruction process to give a single motion free PET/CT volume. All methods were applied to real patient data. The results were evaluated with respect to two quantitative criteria: a) Motion of the left heart ventricle which to assess the accuracy of the motion correction. b) Myocardial wall thickness to assess the quality of images with respect to motion induced blur.
Results: Motion of the heart was reduced to ~0.3 mm. The myocardial wall thickness on the motion corrected data corresponded with that in a single (motionless) respiratory phase. The difference being 0+-0.5mm. Whereas, the difference with the non-motion-corrected data was up to 3mm.
Conclusions: A complete motion correction scheme is evaluated on 7 patient datasets. Both criteria used for quantification show that the respiratory motion was successfully corrected in all cases.
- Society of Nuclear Medicine, Inc.