Dynamic scanning of 15O-butanol with positron emission tomography can identify regional cerebral activations.

To determine task-specific activations of the human brain in individual subjects, we applied pixel-by-pixel t-map statistics to the regional cerebral perfusion data obtained sequentially by dynamic scanning of [15O]-butanol with positron emission tomography (PET). The listmode data were binned into frames of 2 sec, and multiple corresponding pixel-by-pixel activation-minus-control subtractions were used for t-map calculation. The subtraction frames covering 10-40 sec after tracer arrival in the brain showed the activation-related increase of regional cerebral perfusion. A mismatch of the activation and control data by 2 sec resulted in a mean error of <5% of the integrated activity increase. To validate these results, we simulated images with a spatial resolution and signal-to-noise ratio equivalent to that of the [15O]-butanol subtraction images. By means of these simulated images, we determined the minimal data requirements for t-map analysis, the degree of spatial correlations in the image matrix, and the distribution of noise in the t-maps. The simulation results provided a measure to estimate the significance of regional cerebral perfusion changes recorded with [15O]-butanol. The location and spatial extent of regional cerebral activations obtained from dynamic data corresponded closely to those obtained with quantitative measurements of regional cerebral blood flow (rCBF). Our results show that statistical parametric mapping of [15O]-butanol scanning data allows the detection of significant, task-specific brain activations in single activation-control comparisons in individual subjects. Copyright (c) 1997 Wiley-Liss, Inc.