Abstract
1934
Objectives Head motion was commonly happened during the period of PET scanning, which would lead to severe artifacts in PET image reconstruction. In this study, a line of response (LOR) based adaptive motion correction method was presented to solve this problem.
Methods The method mainly contains three important steps. First, without introducing additional motion tracking devices, the acquired data was dynamically reconstructed with very short time frames to find out the time intervals during which the motion happened. Second, from the multiple dynamic frames, head motion model could be established by mutual information based three-dimensional rigid registration method. Third, an event-by-event motion correction was performed on the list-mode data. The performance of the method was evaluated by several clinical datasets. All the datasets were obtained from uMI510 produced by United Imaging Healthcare. In the first experiment, two scans of the same patient were acquired. The patient was with frequent head motion in the first scan and kept steady in the second scan. Then the misplaced LOR was corrected into the right position to reconstruct the image. To further demonstrate the feasibly of the method, another two datasets were used. In one scan the patient was frequently shaking, and in the other scan the patient was frequently nodding. Then we compared their data before and after correction with this method.
Results Results of the first experiment showed that the proposed method could well detect the time intervals during which the motion happened. Then, moved LORs were corrected into their right position and the images were reconstructed. After motion correction, the resultant images were comparable to those from non-motion data. There were a lot of motion artifacts before correction. The images were blurry and the details cannot be showed. After correction the brain tissue structure of the images displayed clearly. There was no significant difference between the corrected images and non-motion images in the second experiment. Both results showed that our method could effectively detect the motion and eliminate the motion artifacts.
Conclusions In this paper, a motion correction method was proposed for PET/CT brain imaging. This method is adaptive, robust and effective, and has been validated by clinical datasets.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.