Determining Significant Within‐Subject Changes of BOLD MRI Cerebrovascular Reactivity to CO2

Abstract

Cerebrovascular reactivity (CVR) is the ratio of the cerebral blood flow (CBF) response to an increase in a vasoactive stimulus. Current measurement of CVR uses the BOLD MRI signal as an indicator of CBF in response to standardized changes in CO2. Despite uniform test conditions, there are still test‐to‐test differences in CVR due to variations in physiology and technology over time. Thus performing longitudinal studies requires the quantification of abnormal differences in CVR over time. We therefore generated a database of normal CVR test‐retest variability as a normal reference to statistically score differences in CVR. Two standardized CVR tests were conducted in fifteen healthy males (36.7 ± 16.1 y) within a three week period. All scans were co‐registered into standard space and a normal difference CVR map was calculated from the two time points for all voxels for each of the 15 subjects consisting of the mean and standard deviation of the differences to generate an interval test difference (ID) atlas. We illustrate our concept by generating ID z‐maps for patients with steno‐occlusive disease who had CVR tests done before and after surgical intervention. Differences in CVR values for the patients were compared to the ID atlas and scored as z‐values, the fractional standard deviation from the corresponding voxel of the atlas. The ID z‐maps confirmed improvement brought about by surgical intervention, and gave an indication of the extent and distribution of changes in CVR. The application of ID z‐maps will enable study of the natural history of cerebrovascular disease and response to treatment.