Data analysis in behavioral cerebral blood flow activation studies using xenon-133 clearance.

Publisher Website
Google Scholar
Abstract
Three mainstream strategies exist to detect the responses of regional cerebral blood flow to functional activation. We tested the significance of changes in raw regional cerebral blood flow data, regional cerebral blood flow data normalized by division by global cerebral blood flow (dependent model of the regional-to-global cerebral blood flow relation), and regional cerebral blood flow data treating global cerebral blood flow as a covariate (independent model). Both latter models attempt to enhance regional sensitivity by removing global effects. We examined the sensitivity and pitfalls of these three strategies in behavioral activation studies. These three strategies of data analysis were applied to changes in regional cerebral blood flow induced by a visuospatial problem-solving task in 38 healthy subjects as measured by the intravenous xenon-133 method with 32 stationary detectors. Mental activation increased blood flow in all regions of interest. Raw data were most sensitive and reliable to detect responses to mental stimulation. Both the independent and dependent models to remove global effects were less sensitive and falsely indicated deactivation in regions that were clearly stimulated. In behavioral activation paradigms, safe data analysis should be restricted to using raw regional cerebral blood flow increases without normalization or separation of global from regional effects. Studies using complex stimulation tasks should be scrutinized for global cerebral blood flow effects confounding regional responses.