Subject motion during a positron emission tomography (PET) study can seriously degrade image quality and affect the quantitative results. A number of methods are in use to restrain patients during brain PET scans; however, all but the most severe schemes are inadequate for completely eliminating head motion. For scanners with high spatial resolution, this motion can be a significant determinant of the achievable spatial resolution in the image. We have investigated the use of a real-time, ancillary measurement of patient head movement to correct for motion distortion in brain PET images. Past work included using phantom studies to demonstrate the feasibility of an on-line rearrangement of the coincidence data. The method suffered from the major drawback that the spacetracking device used was not accurate in the metal environment of the PET scanner. This year, we have developed a method to measure patient motion which can be operated while the patient is in the scanner minimal patient disturbance. The device is currently undergoing tests in normal volunteers against a highly accurate spacetracking device to confirm its accuracy and reliability. The final stages of the project are two-fold: (1) validation of the device in phantoms and patients undergoing scans in the Scanditronix/GE PC2048-15B brain PET scanner with post-acquisition correction of the image data and (2) development of a real-time correction for patient motion as the data are acquired. The first phase represents a path towards rapid implementation of motion correction in routine PET studies. The second is a continuation of the earlier feasibility study.
