Elsevier

NeuroImage

Volume 39, Issue 1, 1 January 2008, Pages 206-214
NeuroImage

Controlled inspiration depth reduces variance in breath-holding-induced BOLD signal

https://doi.org/10.1016/j.neuroimage.2007.08.014Get rights and content

Abstract

Recent studies have shown that blood oxygen level dependent (BOLD) response amplitude during short periods of breath holding (BH) measured by functional magnetic resonance imaging (fMRI) can be an effective metric for intersubject calibration procedures. However, inconsistency in the depth of inspiration during the BH scan may account for a portion of BOLD variation observed in such scans, and it is likely to reduce the effectiveness of the calibration measurement. While modulation of BOLD signal has been correlated with end-tidal CO2 and other measures of breathing, fluctuations in performance of BH have not been studied in the context of their impact on BOLD signal. Here, we studied the degree to which inspiration depth corresponds to BOLD signal change and tested the effectiveness of a method designed to control inspiration level through visual cues during the BH task paradigm. We observed reliable differences in BOLD signal amplitude corresponding to the depth of inspiration. It was determined that variance in BOLD signal response to BH could be significantly reduced when subjects were given visual feedback during task inspiration periods. The implications of these findings for routine BH studies of BOLD-derived neurovascular response are discussed.

Introduction

One of the primary methodological challenges facing fMRI is that the blood oxygen level dependent (BOLD) signal is only an indirect measure of neuronal activity because it derives from neurovascular processes (Ogawa et al., 1990). There is concern that inferences regarding cerebral activation may be confounded by regional variation in brain vasculature and by differences in the cerebral vascular system amongst subjects or between different subject groups. For example, acute ingestion of caffeine is known to reduce cerebral blood flow (CBF) in healthy volunteers (Mulderink et al., 2002). Moreover, several studies have documented age-related decline in neurovascular reactivity and elasticity with age (Hajdu et al., 1990, Kastrup et al., 1998a, Mann et al., 1986, Reich and Rusinek, 1989), while another showed greater proportions of noise in children’s BOLD response relative to that in adults (Thomason et al., 2005). Additionally, studies of global BOLD response in children and adults have demonstrated variations in BOLD response across different brain regions (Kastrup et al., 1999a, Thomason et al., 2005). Variations in BOLD response may be related to differences in baseline cerebral blood flow and/or volume, to vascular composition (Jennings et al., 1998), or even to rate of breathing or heart rate (Birn et al., 2006). Thus, inferences drawn from BOLD imaging will benefit from methods developed to better discriminate fMRI signal components related to neural activity from those that result from intrinsic properties of the local vasculature. The coupling of neural events to BOLD signal can be tightened if the neurovascular response is obtained routinely during scanning and included in signal analysis (Bandettini and Wong, 1997, Cohen et al., 2004, Handwerker et al., 2007, Thomason et al., 2007).

At present, there are two predominant ways to query global, hypercapnia-based neurovascular response in a human fMRI session, CO2/O2 inhalation (Bandettini and Wong, 1997, Cohen et al., 2004, Cohen et al., 2002, Corfield et al., 2001, Davis et al., 1998, Reich and Rusinek, 1989, Rostrup et al., 2000, Schwarzbauer and Heinke, 1999, Schwarzbauer and Hoehn, 2000, Vazquez et al., 2006) and breath holding (BH) (Kastrup et al., 1999a, Kastrup et al., 1999b, Kastrup et al., 1998b, Li et al., 2000, Li et al., 1999a, Li et al., 1999b, Liu et al., 2002, Nakada et al., 2001, Riecker et al., 2003, Stillman et al., 1995, Thomason et al., 2005). In 2001, Kastrup and colleagues (2001) demonstrated cerebrovascular reactivity resulting from both techniques was similar. BH does not require use of tight-fitting masks to deliver air-mixtures and is non-invasive, giving it a natural advantage for routine use over methods requiring the administration of CO2 or O2 during a scanning session. Furthermore, the process of collecting BH data is similar to that of scanning during cognitive tasks from the standpoint of the subject and requires acquisition parameters that are typical in most fMRI studies of human cognition. Much like a cognitive paradigm, the timing of the holding and the regular-breathing periods is indicated by visual cues. Additionally, the neurovascular calibration data and the cognitive task of interest may be collected with the same slice prescription and other scan parameters. Already studies have demonstrated good compliance by elderly subjects and children to short periods (∼ 20 s) of BH for the express purpose of measuring neurovascular response during BOLD fMRI (Riecker et al., 2003, Thomason et al., 2005).

The BH task can be carried out either by holding one’s breath after inspiration or after expiration. Both have proven to be effective methods for measuring neurovascular reactivity and the measurements derived from these methods have proven to be comparable (Kastrup et al., 1999b, Kastrup et al., 1998b, Leoni et al., 2007). One important difference, however, is end-inspiration (holding one’s breath after inspiration), is easier to perform and thus may be easier for a wide range of subjects to perform, particularly aged or infirm volunteers. In their study, Kastrup and colleagues (1998b) used shorter periods of holding breath following expiration in order to counter this difference between the tasks.

BH-induced BOLD calibration is an emergent method for correction of vascular reactivity-induced effects in the BOLD signal. Researchers are using measurements taken during BH to identify individual- and region-specific differences in hemodynamic responsivity, and applying correction for these differences to cognitive paradigms. Handwerker and colleagues (2007) used end-expiration BH data to calibrate individual subject activations during a visuomotor saccade task. Thomason and colleagues used end-inspiration BH data to calibrate individual subject activations during a working memory task and reported that BH-derived calibration data applied on a voxel-wise basis effectively reduced intersubject variability by 24.8% (Thomason et al., 2007). In both studies, the hypercapnia tasks accounted for a significant amount of the BOLD signal variability observed during the cognitive tasks.

Past studies of BH-induced BOLD signal change have focused on duration of hold rather than depth of inspiration or fullness of expiration. Recent work has drawn attention to the influence of breath volume and rate on BOLD signal change. Wise and colleagues (2004) have described correlations between resting state BOLD signal and changes in end-tidal CO2 (etCO2) levels and breathing states. They demonstrated BOLD signal is affected by breath inspiration level, and in that study the authors recommended caution in interpreting results in designs that alter breathing conditions. More recent results from the same group (Wise et al., 2007) suggested that feedback based on etCO2 levels could provide improved precision in mapping of CO2 reactivity. In addition, Birn and colleagues (2006) recently demonstrated that significant variance in resting state BOLD response was accounted for by breathing volume patterns, suggesting the sensitivity of BOLD signal to the breathing patterns, themselves. Therefore, we hypothesize that active control of the BH inspiration level will lead to improved mapping of BOLD BH-induced percent signal change.

The goals of the present study were to test the hypothesis that inspiration depth will impact BOLD signal amplitude, and to examine whether a paradigm that includes feedback cueing subjects to inspire to a controlled depth is a significant improvement over BH paradigms without feedback. The results of this study may help to optimize methods for collecting BH data for routine use in scan protocols.

Section snippets

Subjects

Data were collected from 13 healthy adults from Stanford University and the surrounding community (6 females, 7 males; mean age 30, range 23 to 64 years) after giving informed consent as approved by the Stanford Institutional Review Board. Because they should have no effect on neurovascular reactivity, handedness and native language were not controlled during subject selection. Data from two subjects were discarded due to poor behavioral performance of the task (did not achieve target

fMRI data

Activation maps were generated for each scan by correlating linearly detrended image time series data with sine and cosine waveforms having period equal to that of the task (30 s). Using both sine and cosine correlates allowed the temporal phase as well as maximum correlation magnitude to be determined without assuming a hemodynamic response model. Because the task blocks were only 15 s long, higher order harmonics were largely attenuated due to filtering by the hemodynamic response. A sigma

Behavioral data

All subjects were able to hold their breath as described by the study protocol. On average, subjects (n = 10) were above 92% accurate (x = 92.25%) across c and v scans at attaining the cued target inspiration set points. xfluc, or variation across trials in holding one’s breath, was significantly different between controlled-feedback (xfluc = 0.062) and no-feedback scans (xfluc = 0.094), t(10) = 3.33, p < 0.01. Variation in breath depth was reduced with controlled feedback.

The BOLD results for the task

Discussion

The approach presented herein extends the precision of the BH method for obtaining rapid, global systemic BOLD calibration data. The present study demonstrated that depth of inspiration affects BOLD signal change amplitude (Fig. 3, Fig. 4). These results were not affected by task-correlated motion, as these components were found to amount to only 0.19 ± 0.13 pixel in-plane. The correspondence between inspiration depth and BOLD signal change was highly significant and is likely related to a

Acknowledgments

We are grateful to several members of the fBIRN consortium, specifically Lee Friedman, Tom Liu and Doug Greve for valuable discussions, to anonymous reviewers for constructive comments, and to the NIH for funding: P41 RR09784 and M01 RR000827.

References (40)

  • E. Rostrup et al.

    Regional differences in the CBF and BOLD responses to hypercapnia: a combined PET and fMRI study

    NeuroImage

    (2000)
  • A.E. Stillman et al.

    Functional MRI of brain during breath holding at 4 T

    Magn. Reson. Imaging

    (1995)
  • M.E. Thomason et al.

    Breath holding reveals differences in fMRI BOLD signal in children and adults

    NeuroImage

    (2005)
  • A.L. Vazquez et al.

    Vascular dynamics and BOLD fMRI: CBF level effects and analysis considerations

    NeuroImage

    (2006)
  • R.G. Wise et al.

    Resting fluctuations in arterial carbon dioxide induce significant low frequency variations in BOLD signal

    NeuroImage

    (2004)
  • P.A. Bandettini et al.

    A hypercapnia-based normalization method for improved spatial localization of human brain activation with fMRI

    NMR Biomed.

    (1997)
  • E.R. Cohen et al.

    Effect of basal conditions on the magnitude and dynamics of the blood oxygenation level-dependent fMRI response

    J. Cereb. Blood Flow Metab.

    (2002)
  • T.L. Davis et al.

    Calibrated functional MRI: mapping the dynamics of oxidative metabolism

    Proc. Natl. Acad. Sci. U. S. A.

    (1998)
  • K.J. Friston et al.

    Movement-related effects in fMRI time-series

    Magn. Reson. Med.

    (1996)
  • G.H. Glover et al.

    Self-navigated spiral fMRI: interleaved versus single-shot

    Magn. Reson. Med.

    (1998)
  • Cited by (55)

    • The Problem of Neurovascular Uncoupling

      2021, Neuroimaging Clinics of North America
      Citation Excerpt :

      Not all patients can tolerate the breathing-mask, and these patients are often excluded from CO2 breathing studies for CVR measurements. Thus, breath-holding (BH) BOLD fMR imaging is becoming a popular alternative for CVR mapping.19,21,41–43,65–75 BH CVR mapping has been demonstrated to be comparable in CVR assessment to gas inhalation techniques71 and is thus considered reliable.66

    • Fourier modeling of the BOLD response to a breath-hold task: Optimization and reproducibility

      2016, NeuroImage
      Citation Excerpt :

      However, noninvasive respiratory tasks have also been proposed, which present the important advantages of not requiring complex experimental setups including uncomfortable breathing masks and being well tolerated by healthy volunteers as well as patients. Most notably, breath-hold (BH) tasks are commonly used in both clinical and research settings, leading to mild hypercapnia and vasodilation and inducing measurable CBF and BOLD signal increases (Bright and Murphy, 2013; Kastrup et al., 2001, 1999; Lipp et al., 2015; Magon et al., 2009; Pillai and Mikulis, 2014; Scouten and Schwarzbauer, 2008; Thomason and Glover, 2008). In contrast to techniques based on the controlled manipulations of respiratory gases, the use of BH tasks does not allow for precise control of end-tidal CO2 pressure (PETCO2) levels and hence hypercapnia levels.

    • The Function Biomedical Informatics Research Network Data Repository

      2016, NeuroImage
      Citation Excerpt :

      Since there is no obvious way to calibrate functional signals from cognitive tasks, FBIRN developed a sensorimotor (SM) task, to activate motor, visual, and auditory cortices, and a breath-hold (BH) task to measure vascular response. The imaging protocol at each site included: high-resolution T1-weighted, T2-weighted, and fMRI scans of Mismatch Negativity, Serial Item Recognition Paradigm (SIRP, 2 runs) (Potkin et al., 2009a), SM (Friedman and Glover, 2006b), resting state (REST), and BH (Thomason and Glover, 2008; Thomason et al., 1999) tasks. All tasks were performed in the same order at each site (see S1 Table 4).

    View all citing articles on Scopus
    View full text