Inter-Subject Variability in Hypercapnic Normalization of the BOLD FMRI Response

Joy Liau¹, Thomas Liu²

Hypercapnic normalization has been proposed as a method for reducing inter-subject variability in the amplitude of the BOLD fMRI signal. Some prior studies have reported a decrease in variability, while others have found an increase. Here we show that the presence of an intercept term in the relation between the functional and hypercapnic BOLD responses results in a systematic bias term in the normalized response that can lead to increased inter-subject variability. An alternate approach that treats the hypercapnic BOLD response as a covariate can be used to minimize this bias term and reduce inter-subject variability in the normalized response.