Quantitative fMRI using hyperoxia calibration: Reproducibility during a cognitive Stroop task

Abstract

Arterial spin labelling allows simultaneous measurement of both the blood-oxygenation-level-dependent (BOLD) and the cerebral blood flow (CBF) response to changes in neural activity. The addition of a hypercapnia or hyperoxia calibration allows additional quantification of changes in the cerebral metabolic rate of oxygen (CMRO₂). In this study we test the reproducibility of measurements derived using the hyperoxia approach, during a cognitive Stroop task. A QUIPSSII sequence is used at 3 T to collect simultaneous CBF and BOLD signal during two 3 min periods of hyperoxia and an 8 min Stroop task. Hyperoxia was administered via an open system and end-tidal values were sampled via a nasal cannula; average end-tidal values of 60% were reached. This procedure is repeated to allow the reproducibility of the estimated parameters to be tested. The use of a cognitive Stroop task allows testing of the measurements in frontal and parietal regions as well as sensorimotor areas in which previous studies have been focussed. We find reduced reproducibility of the calculated parameters compared to the hypercapnia approach, thought to be attributable to lower absolute BOLD and CBF responses. In particular we do not find ‘n’ to have improved reproducibility compared to other parameters, as has been found in previous work using the hypercapnia approach. Across all brain areas we report a value of ΔCMRO₂ of 12% and neurovascular coupling constant n of 2.5. Interestingly we find n to be higher in parietal and frontal areas in comparison to the primary motor cortex.

Introduction

The Blood-Oxygenation-Level-Dependent (BOLD) signal is widely used in studies of normal brain function, but clinical applications are problematic due to possible alterations in neurovascular coupling, making signal interpretation difficult (Pineiro and Pendlebury, 2002, D'Esposito and Deouell, 2003, Detre, 2006, Jezzard and Buxton, 2006). The BOLD signal is sensitive to the deoxyhaemoglobin content of the blood, which is altered during changes in neural activity due to concomitant changes in cerebral blood flow (CBF), cerebral blood volume (CBV) and cerebral metabolic rate of oxygen (CMRO₂) (Buxton et al., 2004). It is possible to separate these physiological contributions by using an arterial spin labelled (ASL) pulse sequence, and including an iso-metabolic calibration scan, where it is assumed that there is no change in oxygen metabolism. The most common approach is to administer approximately 5% CO₂ in order to induce hypercapnia (Davis et al., 1998), however some concern remains as to whether this approach is truly iso-metabolic (Kliefoth and Grubb, 1979, Jones and Berwick, 2005, Sicard and Duong, 2005, Zappe and Uludag, 2008). Recently, a new approach to calibration using hyperoxia has been suggested which offers a number of possible benefits, including increased comfort for the participant, and no reliance on the noisy ASL signal for calibration (Chiarelli et al., 2007a).

Most previous work using a calibration approach for quantitative fMRI has considered activation of primary visual and motor cortices (Davis and Kwong, 1998, Chiarelli and Bulte, 2007b, Leontiev and Buxton, 2007) with only very recent work concerning cognitive tasks (Restom and Perthen, 2008, Restom and Bangen, 2007). If this approach is to become used more widely for clinical applications it is important to test the reproducibility of the measurements for cognitive fMRI paradigms, often involving smaller BOLD activations than in visual or motor cortices. The aim of our work is to test the reproducibility of the measurements derived from the hyperoxia calibration approach for a cognitive Stroop task (Zysset et al., 2001). We use a reduced hyperoxia protocol taking only 13 min which also makes this approach more clinically feasible than previous work (Chiarelli et al., 2007a). In this study, we also compare quantitative fMRI measures across activated brain regions.