Head movement in normal subjects during simulated PET brain imaging with and without head restraint

Head movement during brain imaging is recognized as a source of image degradation in PET and most other forms of medical brain imaging. However, little quantitative information is available on the kind and amount of head movement that actually occurs during these studies. We sought to obtain this information by measuring head movement in normal volunteers.

Head position data were acquired for 40 min in each of 13 supine subjects with and without head restraint. These data were then used to drive a mathematically simulated head through exactly the same set of movements. The positions of point sources embedded in this head were computed at each location and these data summarized as movement at FWHM in each of the three coordinate directions.

Head movement increased with the length of the sampling interval for studies of either type (with or without head restraint), but the amount and rate of increase with restraint was much smaller. In contrast, head movement during consecutive, short sampling intervals was small and did not increase with time. Spatial gradients in head movement were detected within each study type, and significant spatial differences in head movement were found between study types.

Head movements in normal, supine subjects, though small, can cause the effective resolution of a brain imaging study to appear to vary in space and time. These effects can be reduced significantly with head restraint and may also be reduced by dividing the acquisition of a single image into a sequence of short images (instead of a single long image), aligning these images spatially and summing the result.