Motion correction of PET images using multiple acquisition frames

Abstract

Positron emission tomography (PET) is a relatively lengthy brain imaging method. Because it is difficult for the subject to stay still during the data acquisition, head motion during scans is a source of image degradation. A simple data acquisition technique to reduce the effect of this problem is described. The technique associates the incoming data with the real-space position of the head. During the PET scan, the head position is constantly monitored with two video cameras and compared to its initial position. Every time the displacement for a region within the field of view (FOV) is larger than a specified threshold displacement, the PET data acquisition system starts to save the PET data in a new frame. The total number of frames required for a complete study depends on the magnitude of the head motion during the study and on the threshold displacement. At the end of the study, all the acquired frames are reconstructed independently and each image is rotated and translated to coincide with the initial position. When these images are summed, they produce a final image with fewer motion artefacts.