Low-frequency respiration related signals in resting state fMRI: a comparison of end-tidal CO2 and respiration volume per time

Introduction Functional connectivity analysis has been used to map several consistent resting state networks by exploiting coherent low-frequency fluctuations in the BOLD signal. The assumption in most of the fMRI literature is that these correlated fluctuations are neuronal in origin. Others sources of low-frequency fluctuations exist in fMRI data that are unrelated to neuronal activity. These must be removed to achieve accurate mapping of resting state networks. Fluctuations in arterial CO2 in the range of 0-0.05Hz are related to changes in the breathed volume of air and are reflected in the BOLD signal. Two readily available surrogate measurements are end-tidal CO2 and RVT (respiration volume over time) [1,2]. Physiological noise correction using end-tidal CO2, a measure of the CO2 concentration at the end of exhalation, and RVT, a measure of breathed volume over time, would be expected to remove similar components from the fMRI signal (although the measures are anti-correlated with each other). Since there is a cost in equipment, setup time and discomfort to the subjects, it is important to investigate the relationship between these measures to remove any redundancy. Accurate correction of this low-frequency noise component will increase the precision of functional connectivity analysis, improving its applicability to both healthy subject and clinical studies.