Department of Radiology, School of Medicine, Johns Hopkins University, Baltimore, USA
With the arrival of increasingly higher resolution PET systems, small amounts of motion can cause significant blurring in the images, compared to the intrinsic resolutions of the scanners. In this work, we have reviewed advanced correction methods for the three cases of (i) unwanted patient motion, as well as motions due to (ii) cardiac and (iii) respiratory cycles. For the first type of motion (most often studies in PET brain imaging), conventional motion-correction algorithms have relied on extraction of the motion information from the emission data itself. However, the accuracy of motion compensation in this approach is degraded by the noisy nature of the emission data. Subsequently, advanced methods, as reviewed in this work, make use of external real-time measurements of motion. Various image-based and projection-based correction methods have been discussed and compared. The paper also reviews recent and novel applications that perform corrections for cardiac and respiratory motions. Unlike conventional gating schemes, in which the cardiac and respiratory gated frames are independently reconstructed (resulting in noisy images), the reviewed methods are seen to follow a common trend of seeking to produce images of higher quality by making collective use of all the gated frames (and the estimated motion). As an observation, a general theme in motion-correction methods is seen to be the use of increasingly sophisticated software to make use of existing advanced hardware. In this sense, this field is very open to future novel ideas (hardware, and especially software) aimed at improving motion detection, characterization and compensation.