Most analysis of activation data in PET and fMRI involves relative changes in rCBF distribution between conditions. This is in part due to the prevailing assumption that pattern changes and not fluctuations in absolute rCBF represent the important data regarding brain activity-cognition relationships. Normalized redistribution analysis invariably shows “deactivation” as well as activation. We present data showing that “deactivation” can represent either true reduction in cerebral activity or be an artifact of normalization that misses important global changes. These changes appear to reflect the extent of novelty versus routinization of task and control conditions.
Methods
27 subjects participated in three protocols involving 7 tasks and baseline scans during quantitative rCBF studies: 1) Passive visual stimulation involving multiple images, followed by a forced choice recognition task, 2) 3-dimensional mental rotation and speech sound discrimination, 3) Linguistic Components - phonetic (“br”) and semantic (“living”) target tasks and continuous recognition memory of a word series. Whole brain scans were performed every 10 seconds using a modified Picker triple-head tomograph. rCBF was calculated from reconstructed Xe-133 clearance data (24 scans); final images were generated representing rCBF in ml/100g/min. Data analysis was performed to determine the pattern of redistribution of rCBF and the absolute rCBF changes associated with significant activation or “deactivation” findings via: 1) an ROI method using ANOVA on a multi-region profile of mean absolute rCBF, 2) SPM2, 3) inspection of quantitative mean rCBF images created using spatially normalized Xe SPECT scans.
Results
All tasks except passive visual stimulation and the semantic target task resulted in absolute increases in all cortical ROI. Only in these two conditions did SPM and absolute rCBF analysis concur regarding “deactivated” regions. Visual stimulation resulted in decreased rCBF in frontal ROI along with increase in occipital cortex. SPM identified activation and “deactivation” consistent with these absolute rCBF changes. There was concordance between SPM “deactivation” and absolute rCBF reduction in right temporal-parietal regions during the semantic task. In all other conditions there was agreement in activated regions, but significant “de-activation” on SPM was seen only as a less-than-average increase in absolute rCBF compared to other regions. This occurred in parietal-temporal areas during the phonetic task and in frontal cortex during auditory memory and mental rotation.
Conclusion
Redistribution of rCBF without change in mean cortical rCBF seems to occur in cases of passive sensory stimulation and in task conditions involving fairly routine or well mastered operations, such as our semantic task in which rCBF was lower compared to other linguistic operations and resting baseline. In such situations there is concordance between deactivation seen on SPM and absolute rCBF analysis. Global increases occur in situations involving more novel cognitive operations, and may include control conditions designed to isolate “lower” stages of processing but which may be less familiar in isolation and thus more contrived. Our data indicate that although regional reductions in absolute activity do occur, at least part of the “deactivation” seen on non-quantitative imaging and typical SPM analysis does not represent physiological reduction because systematic global changes are ignored.