Development of motion correction technique for cardiac ¹⁵O-water PET study using an optical motion tracking system

Objective: Cardiac ¹⁵O-water PET studies provide an accurate quantitation of regional myocardial blood flow (rMBF). We developed a motion correction system using an optical motion-tracking device to detect a subject's global movement for cardiac study. Methods: PET studies were carried out on a cardiac phantom and a healthy volunteer at rest. The three-dimensional locations of the markers attached to the subjects during scans were measured using an optical motion-tracking system. In the phantom study, we performed a transmission scan and seven ¹⁸F emission scans of a baseline and with artificial misalignment of shifts and rotations. The correlation coefficients between the baseline and the other images before and after the corrections for the misalignment were calculated. In the human study, we performed a ¹⁵O-water dynamic scan with a transmission and axially 30 mm-shifted transmission scans. Motion of the subject was estimated by the information from the system, and was corrected on each sinogram using attenuation maps realigned to dynamic frames. Reconstructed dynamic images were then realigned to the transmission data. We calculated rMBF values for nine segments and myocardial images from the emission images, which were reconstructed with the first attenuation map (reference) and with the misaligned attenuation map before and after our corrections. Results: In the phantom study, the correlation coefficients were improved from 0.929 ± 0.022 to 0.987 ± 0.010 (mean ± SD) after the corrections. In the human study, the global and cyclic movements were detected. The cyclic movement due to respiration was smoothed by frame-averaging, and reasonable information of the global movement was obtained. The rMBF value (mean ± SD) was 0.94 ± 0.12 mL/min/g for the reference. The rMBF values using the misaligned attenuation map changed from 1.03 ± 0.21 to 0.93 ± 0.11 mL/min/g after the correction, and spurious defects in myocardial images were also recovered. Conclusions: Our technique provided reasonable information for correcting the global movement of the subject. It was shown that this system was applicable to detect and correct subject movement in cardiac PET studies at rest.