Reliable quantification of BOLD fMRI cerebrovascular reactivity despite poor breath-hold performance

Potential pitfalls when denoising resting state fMRI data using nuisance regression

In resting state fMRI, it is necessary to remove signal variance associated with noise sources, leaving cleaned fMRI time-series that more accurately reflect the underlying intrinsic brain fluctuations of interest. This is commonly achieved through nuisance regression, in which the fit is calculated of a noise model of head motion and physiological processes to the fMRI data in a General Linear Model, and the "cleaned" residuals of this fit are used in further analysis. We examine the statistical assumptions and requirements of the General Linear Model, and whether these are met during nuisance regression of resting state fMRI data. Using toy examples and real data we show how pre-whitening, temporal filtering and temporal shifting of regressors impact model fit. Based on our own observations, existing literature, and statistical theory, we make the following recommendations when employing nuisance regression: pre-whitening should be applied to achieve valid statistical inference of the noise model fit parameters; temporal filtering should be incorporated into the noise model to best account for changes in degrees of freedom; temporal shifting of regressors, although merited, should be achieved via optimisation and validation of a single temporal shift. We encourage all readers to make simple, practical changes to their fMRI denoising pipeline, and to regularly assess the appropriateness of the noise model used. By negotiating the potential pitfalls described in this paper, and by clearly reporting the details of nuisance regression in future manuscripts, we hope that the field will achieve more accurate and precise noise models for cleaning the resting state fMRI time-series.

Time delay processing of hypercapnic fMRI allows quantitative parameterization of cerebrovascular reactivity and blood flow delays

Blood oxygenation level-dependent fMRI contrast depends on the volume and oxygenation of blood flowing through the circulatory system. The effects on image intensity depend temporally on the arrival of blood within a voxel, and signal can be monitored during the time course of such blood flow. It has been previously shown that the passage of global endogenous variations in blood volume and oxygenation can be tracked as blood passes through the brain by determining the strength and peak time lag of their cross-correlation with blood oxygenation level-dependent data. By manipulating blood composition using transient hypercarbia and hyperoxia, we can induce much larger oxygenation and volume changes in the blood oxygenation level-dependent signal than result from natural endogenous fluctuations. This technique was used to examine cerebrovascular parameters in healthy subjects (n = 8) and subjects with intracranial stenosis (n = 22), with a subgroup of intracranial stenosis subjects scanned before and after surgical revascularization (n = 6). The halfwidth of cross-correlation lag times in the brain was larger in IC stenosis subjects (21.21 ± 14.22 s) than in healthy control subjects (8.03 ± 3.67), p < 0.001, and was subsequently reduced in regions that co-localized with surgical revascularization. These data show that blood circulatory timing can be measured robustly and longitudinally throughout the brain using simple respiratory challenges.

Optical measures of changes in cerebral vascular tone during voluntary breath holding and a Sternberg memory task

The human cerebral vasculature responds to changes in blood pressure and demands for oxygenation via cerebral autoregulation. Changes in cerebrovascular tone (vasoconstriction and vasodilation) also mediate the changes in blood flow measured by the BOLD fMRI signal. This cerebrovascular reactivity is known to vary with age. In two experiments, we demonstrate that cerebral pulse parameters measured using optical imaging can quantify changes in cerebral vascular tone, both globally and locally. In experiment 1, 51 older adults (age range=55-87) performed a voluntary breath-holding task while cerebral pulse amplitude measures were taken. We found significant pulse amplitude variations across breath-holding periods, indicating vasodilation during, and vasoconstriction after breath holding. The breath-holding index (BHI), a measure of cerebrovascular reactivity (CVR) was derived and found to correlate with age. BHI was also correlated with performance in the Modified Mini-Mental Status Examination, even after controlling for age and education. In experiment 2, the same participants performed a Sternberg task, and changes in regional pulse amplitude between high (set-size 6) and low (set-size 2) task loads were compared. Only task-related areas in the fronto-parietal network (FPN) showed significant reduction in pulse amplitude, indicating vasodilation. Non-task-related areas such as the somatosensory and auditory cortices did not show such reductions. Taken together, these experiments suggest that optical pulse parameters can index changes in brain vascular tone both globally and locally, using both physiological and cognitive load manipulations.

Hypercapnic evaluation of vascular reactivity in healthy aging and acute stroke via functional MRI

Functional MRI (fMRI) is well-established for the study of brain function in healthy populations, although its clinical application has proven more challenging. Specifically, cerebrovascular reactivity (CVR), which allows the assessment of the vascular response that serves as the basis for fMRI, has been shown to be reduced in healthy aging as well as in a range of diseases, including chronic stroke. However, the timing of when this occurs relative to the stroke event is unclear. We used a breath-hold fMRI task to evaluate CVR across gray matter in a group of acute stroke patients (< 10 days from stroke; N = 22) to address this question. These estimates were compared with those from both age-matched (N = 22) and younger (N = 22) healthy controls. As expected, young controls had the greatest mean CVR, as indicated by magnitude and extent of fMRI activation; however, stroke patients did not differ from age-matched controls. Moreover, the ipsilesional and contralesional hemispheres of stroke patients did not differ with respect to any of these measures. These findings suggest that fMRI remains a valid tool within the first few days of a stroke, particularly for group fMRI studies in which findings are compared with healthy subjects of similar age. However, given the relatively high variability in CVR observed in our stroke sample, caution is warranted when interpreting fMRI data from individual patients or a small cohort. We conclude that a breath-hold task can be a useful addition to functional imaging protocols for stroke patients.

•

Breath-holding can be used to assess the validity of fMRI in stroke patients.

•

Vascular reactivity, estimated by breath-hold fMRI, was greatest in young controls.

•

Acute stroke patients and age-matched controls had similar vascular reactivity.

•

Modeling the breath-hold response on an individual basis can improve results.

The BOLD cerebrovascular reactivity response to progressive hypercapnia in young and elderly

Blood Oxygenation Level Dependent (BOLD) imaging in combination with vasoactive stimuli can be used to probe cerebrovascular reactivity (CVR). Characterizing the healthy, age-related changes in the BOLD-CVR response can provide a reference point from which to distinguish abnormal CVR from the otherwise normal effects of ageing. Using a computer controlled gas delivery system, we examine differences in BOLD-CVR response to progressive hypercapnia between 16 young (28±3years, 9 female) and 30 elderly subjects (66±4years, 13 female). Furthermore, we incorporate baseline T₂* information to broaden our interpretation of the BOLD-CVR response. Significant age-related differences were observed. Grey matter CVR at 7mmHg above resting PetCO₂ was lower amongst elderly (0.19±0.06%ΔBOLD/mmHg) as compared to young subjects (0.26±0.07%ΔBOLD/mmHg). White matter CVR at 7mmHg above baseline PetCO₂ showed no significant difference between young (0.04±0.02%ΔBOLD/mmHg) and elderly subjects (0.05±0.03%ΔBOLD/mmHg). We saw no significant differences in the BOLD signal response to progressive hypercapnia between male and female subjects in either grey or white matter. The observed differences in the healthy BOLD-CVR response could be explained by age-related changes in vascular mechanical properties.

Respiratory challenge MRI: Practical aspects

Respiratory challenge MRI is the modification of arterial oxygen (PaO₂) and/or carbon dioxide (PaCO₂) concentration to induce a change in cerebral function or metabolism which is then measured by MRI. Alterations in arterial gas concentrations can lead to profound changes in cerebral haemodynamics which can be studied using a variety of MRI sequences. Whilst such experiments may provide a wealth of information, conducting them can be complex and challenging. In this paper we review the rationale for respiratory challenge MRI including the effects of oxygen and carbon dioxide on the cerebral circulation. We also discuss the planning, equipment, monitoring and techniques that have been used to undertake these experiments. We finally propose some recommendations in this evolving area for conducting these experiments to enhance data quality and comparison between techniques.

•

Oxygen and carbon dioxide affect cerebral blood flow and metabolism.

•

This can be imaged with various MRI sequences.

•

The practicalities of these techniques are reviewed.

•

Examples of how this has been used to understand disease mechanisms.

Blood oxygen-level dependent functional assessment of cerebrovascular reactivity: Feasibility for intraoperative 3 Tesla MRI

PURPOSE

To assess the feasibility of functional blood oxygen-level dependent (BOLD) MRI to evaluate intraoperative cerebrovascular reactivity (CVR) at 3 Tesla field strength.

METHODS

Ten consecutive neurosurgical subjects scheduled for a clinical intraoperative MRI examination were enrolled in this study. In addition to the clinical protocol a BOLD sequence was implemented with three cycles of 44 s apnea to calculate CVR values on a voxel-by-voxel basis throughout the brain. The CVR range was then color-coded and superimposed on an anatomical volume to create high spatial resolution CVR maps.

RESULTS

Ten subjects (mean age 34.8 ± 13.4; 2 females) uneventfully underwent the intraoperative BOLD protocol, with no complications occurring. Whole-brain CVR for all subjects was (mean ± SD) 0.69 ± 0.42, whereas CVR was markedly higher for tumor subjects as compared to vascular subjects, 0.81 ± 0.44 versus 0.33 ± 0.10, respectively. Furthermore, color-coded functional maps could be robustly interpreted for a whole-brain assessment of CVR.

CONCLUSION

We demonstrate that intraoperative BOLD MRI is feasible in creating functional maps to assess cerebrovascular reactivity throughout the brain in subjects undergoing a neurosurgical procedure. Magn Reson Med 77:806-813, 2017. © 2016 International Society for Magnetic Resonance in Medicine.

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

Cerebrovascular reactivity (CVR) reflects the capacity of blood vessels to adjust their caliber in order to maintain a steady supply of brain perfusion, and it may provide a sensitive disease biomarker. Measurement of the blood oxygen level dependent (BOLD) response to a hypercapnia-inducing breath-hold (BH) task has been frequently used to map CVR noninvasively using functional magnetic resonance imaging (fMRI). However, the best modeling approach for the accurate quantification of CVR maps remains an open issue. Here, we compare and optimize Fourier models of the BOLD response to a BH task with a preparatory inspiration, and assess the test-retest reproducibility of the associated CVR measurements, in a group of 10 healthy volunteers studied over two fMRI sessions. Linear combinations of sine-cosine pairs at the BH task frequency and its successive harmonics were added sequentially in a nested models approach, and were compared in terms of the adjusted coefficient of determination and corresponding variance explained (VE) of the BOLD signal, as well as the number of voxels exhibiting significant BOLD responses, the estimated CVR values, and their test-retest reproducibility. The brain average VE increased significantly with the Fourier model order, up to the 3rd order. However, the number of responsive voxels increased significantly only up to the 2nd order, and started to decrease from the 3rd order onwards. Moreover, no significant relative underestimation of CVR values was observed beyond the 2nd order. Hence, the 2nd order model was concluded to be the optimal choice for the studied paradigm. This model also yielded the best test-retest reproducibility results, with intra-subject coefficients of variation of 12 and 16% and an intra-class correlation coefficient of 0.74. In conclusion, our results indicate that a Fourier series set consisting of a sine-cosine pair at the BH task frequency and its two harmonics is a suitable model for BOLD-fMRI CVR measurements based on a BH task with preparatory inspiration, yielding robust estimates of this important physiological parameter.

Fine tuning breath-hold-based cerebrovascular reactivity analysis models

INTRODUCTION

We elaborate on existing analysis methods for breath-hold (BH)-derived cerebrovascular reactivity (CVR) measurements and describe novel insights and models toward more exact CVR interpretation.

METHODS

Five blood-oxygen-level-dependent (BOLD) fMRI datasets of neurovascular patients with unilateral hemispheric hemodynamic impairment were used to test various BH CVR analysis methods. Temporal lag (phase), percent BOLD signal change (CVR), and explained variance (coherence) maps were calculated using three different sine models and two novel "Optimal Signal" model-free methods based on the unaffected hemisphere and the sagittal sinus fMRI signal time series, respectively.

RESULTS

All models showed significant differences in CVR and coherence between the affected-hemodynamic impaired-and unaffected hemisphere. Voxel-wise phase determination significantly increases CVR (0.60 ± 0.18 vs. 0.82 ± 0.27; P < 0.05). Incorporating different durations of breath hold and resting period in one sine model (two-task) did increase coherence in the unaffected hemisphere, as well as eliminating negative phase commonly obtained by one-task frequency models. The novel model-free "optimal signal" methods both explained the BOLD MR data similar to the two task sine model.

CONCLUSIONS

Our CVR analysis demonstrates an improved CVR and coherence after implementation of voxel-wise phase and frequency adjustment. The novel "optimal signal" methods provide a robust and feasible alternative to the sine models, as both are model-free and independent of compliance. Here, the sagittal sinus model may be advantageous, as it is independent of hemispheric CVR impairment.

Is fMRI "noise" really noise? Resting state nuisance regressors remove variance with network structure

Noise correction is a critical step towards accurate mapping of resting state BOLD fMRI connectivity. Noise sources related to head motion or physiology are typically modelled by nuisance regressors, and a generalised linear model is applied to regress out the associated signal variance. In this study, we use independent component analysis (ICA) to characterise the data variance typically discarded in this pre-processing stage in a cohort of 12 healthy volunteers. The signal variance removed by 24, 12, 6, or only 3 head motion parameters demonstrated network structure typically associated with functional connectivity, and certain networks were discernable in the variance extracted by as few as 2 physiologic regressors. Simulated nuisance regressors, unrelated to the true data noise, also removed variance with network structure, indicating that any group of regressors that randomly sample variance may remove highly structured “signal” as well as “noise.” Furthermore, to support this we demonstrate that random sampling of the original data variance continues to exhibit robust network structure, even when as few as 10% of the original volumes are considered. Finally, we examine the diminishing returns of increasing the number of nuisance regressors used in pre-processing, showing that excessive use of motion regressors may do little better than chance in removing variance within a functional network. It remains an open challenge to understand the balance between the benefits and confounds of noise correction using nuisance regressors.

•

Data variance removed by nuisance regressors contains network structure.

•

Simulated regressors unrelated to noise also extract data with network structure.

•

Random sampling of original data (as few as 10% of volumes) reveals robust networks.

•

After optimal number, motion regressors remove similar variance as simulated ones.

•

Excessive nuisance regressors extract random signal variance with network structure.

Agreement and repeatability of vascular reactivity estimates based on a breath-hold task and a resting state scan

FMRI BOLD responses to changes in neural activity are influenced by the reactivity of the vasculature. By complementing a task-related BOLD acquisition with a vascular reactivity measure obtained through breath-holding or hypercapnia, this unwanted variance can be statistically reduced in the BOLD responses of interest. Recently, it has been suggested that vascular reactivity can also be estimated using a resting state scan.

This study aimed to compare three breath-hold based analysis approaches (block design, sine–cosine regressor and CO₂ regressor) and a resting state approach (CO₂ regressor) to measure vascular reactivity. We tested BOLD variance explained by the model and repeatability of the measures. Fifteen healthy participants underwent a breath-hold task and a resting state scan with end-tidal CO₂ being recorded during both. Vascular reactivity was defined as CO₂-related BOLD percent signal change/mm Hg change in CO₂.

Maps and regional vascular reactivity estimates showed high repeatability when the breath-hold task was used. Repeatability and variance explained by the CO₂ trace regressor were lower for the resting state data based approach, which resulted in highly variable measures of vascular reactivity.

We conclude that breath-hold based vascular reactivity estimations are more repeatable than resting-based estimates, and that there are limitations with replacing breath-hold scans by resting state scans for vascular reactivity assessment.

Measuring vascular reactivity with breath-holds after stroke: a method to aid interpretation of group-level BOLD signal changes in longitudinal fMRI studies

Blood oxygenation level‐dependent (BOLD) contrast functional magnetic resonance imaging (fMRI) is a widely used technique to map brain function, and to monitor its recovery after stroke. Since stroke has a vascular etiology, the neurovascular coupling between cerebral blood flow and neural activity may be altered, resulting in uncertainties when interpreting longitudinal BOLD signal changes. The purpose of this study was to demonstrate the feasibility of using a recently validated breath‐hold task in patients with stroke, both to assess group level changes in cerebrovascular reactivity (CVR) and to determine if alterations in regional CVR over time will adversely affect interpretation of task‐related BOLD signal changes. Three methods of analyzing the breath‐hold data were evaluated. The CVR measures were compared over healthy tissue, infarcted tissue and the peri‐infarct tissue, both sub‐acutely (∼2 weeks) and chronically (∼4 months). In this cohort, a lack of CVR differences in healthy tissue between the patients and controls indicates that any group level BOLD signal change observed in these regions over time is unlikely to be related to vascular alterations. CVR was reduced in the peri‐infarct tissue but remained unchanged over time. Therefore, although a lack of activation in this region compared with the controls may be confounded by a reduced CVR, longitudinal group‐level BOLD changes may be more confidently attributed to neural activity changes in this cohort. By including this breath‐hold‐based CVR assessment protocol in future studies of stroke recovery, researchers can be more assured that longitudinal changes in BOLD signal reflect true alterations in neural activity. Hum Brain Mapp 36:1755–1771, 2015. © 2015 The Authors Human Brain Mapping Published by Wiley Periodicals, Inc.

Comparing cerebrovascular reactivity measured using BOLD and cerebral blood flow MRI: The effect of basal vascular tension on vasodilatory and vasoconstrictive reactivity

Cerebrovascular reactivity (CVR) is an important metric of cerebrovascular health. While the BOLD fMRI method in conjunction with carbon-dioxide (CO2) based vascular manipulation has been the most commonly used, the BOLD signal is not a direct measure of vascular changes, and the use of arterial-spin labeling (ASL) cerebral blood flow (CBF) imaging is increasingly advocated. Nonetheless, given the differing dependencies of BOLD and CBF on vascular baseline conditions and the diverse CO2 manipulation types currently used in the literature, knowledge of potential biases introduced by each technique is critical for the interpretation of CVR measurements. In this work, we use simultaneous BOLD-CBF acquisitions during both vasodilatory (hypercapnic) and vasoconstrictive (hypocapnic) stimuli to measure CVR. We further imposed different levels of baseline vascular tension by inducing hypercapnic and hypocapnic baselines, separately from normocapnia by 4mmHg. We saw significant and diverse dependencies on vascular stimulus and baseline condition in both BOLD and CBF CVR measurements: (i) BOLD-based CVR is more sensitive to basal vascular tension than CBF-based CVR; (ii) the use of a combination of vasodilatory and vasoconstrictive stimuli maximizes the sensitivity of CBF-based CVR to vascular tension changes; (iii) the BOLD and CBF vascular response delays are both significantly lengthened at predilated baseline. As vascular tension can often be altered by potential pathology, our findings are important considerations when interpreting CVR measurements in health and disease.

Factors affecting the determination of cerebrovascular reactivity

BACKGROUND AND PURPOSE

Cerebrovascular reactivity (CVR), measures the ability of the cerebrovasculature to respond to vasoactive stimuli such as CO2. CVR is often expressed as the ratio of cerebral blood flow change to CO2 change. We examine several factors affecting this measurement: blood pressure, stimulus pattern, response analysis and subject position.

METHODS

Step and ramp increases in CO2 were implemented in nine subjects, seated and supine. Middle cerebral artery blood flow velocity (MCAv), and mean arterial pressure (MAP) were determined breath-by-breath. Cerebrovascular conductance (MCAc) was estimated as MCAv/MAP. CVR was calculated from both the relative and absolute measures of MCAc and MCAv responses.

RESULTS

MAP increased with CO2 in some subjects so that relative CVR calculated from conductance responses were less than those calculated from CVR calculated from velocity responses. CVR measured from step responses were affected by the response dynamics, and were less than those calculated from CVR measured from ramp responses. Subject position did not affect CVR.

CONCLUSIONS

(1) MAP increases with CO2 and acts as a confounding factor for CVR measurement; (2) CVR depends on the stimulus pattern used; (3) CVR did not differ from the sitting versus supine in these experiments; (4) CVR calculated from absolute changes of MCAv was less than that calculated from relative changes.

Test-retest reliabilities of resting-state FMRI measurements in human brain functional connectomics: a systems neuroscience perspective

Resting-state functional magnetic resonance imaging (RFMRI) enables researchers to monitor fluctuations in the spontaneous brain activities of thousands of regions in the human brain simultaneously, representing a popular tool for macro-scale functional connectomics to characterize normal brain function, mind-brain associations, and the various disorders. However, the test-retest reliability of RFMRI remains largely unknown. We review previously published papers on the test-retest reliability of voxel-wise metrics and conduct a meta-summary reliability analysis of seven common brain networks. This analysis revealed that the heteromodal associative (default, control, and attention) networks were mostly reliable across the seven networks. Regarding examined metrics, independent component analysis with dual regression, local functional homogeneity and functional homotopic connectivity were the three mostly reliable RFMRI metrics. These observations can guide the use of reliable metrics and further improvement of test-retest reliability for other metics in functional connectomics. We discuss the main issues with low reliability related to sub-optimal design and the choice of data processing options. Future research should use large-sample test-retest data to rectify both the within-subject and between-subject variability of RFMRI measurements and accelerate the application of functional connectomics.

Cerebrovascular reactivity mapping: an evolving standard for clinical functional imaging

SUMMARY

This review article explains the methodology of breath-hold cerebrovascular reactivity mapping, both in terms of acquisition and analysis, and reviews applications of this method to presurgical mapping, particularly with respect to blood oxygen level-dependent fMRI. Its main application in clinical fMRI is for the assessment of neurovascular uncoupling potential. Neurovascular uncoupling is potentially a major limitation of clinical fMRI, particularly in the setting of mass lesions in the brain such as brain tumors and intracranial vascular malformations that are associated with alterations in regional hemodynamics on either an acquired or congenital basis. As such, breath-hold cerebrovascular reactivity mapping constitutes an essential component of quality control analysis in clinical fMRI, particularly when performed for presurgical mapping of eloquent cortex. Exogenous carbon dioxide challenges used for cerebrovascular reactivity mapping will also be discussed, and their applications to the evaluation of cerebrovascular reserve and cerebrovascular disease will be described.

Cerebrovascular reactivity in the brain white matter: magnitude, temporal characteristics, and age effects

White matter (WM) comprises about half of the brain and its dysfunction is implicated in many brain disorders. While structural properties in healthy and diseased WM have been extensively studied, relatively little is known about the physiology underlying these structural characteristics. Recent advances in magnetic resonance (MR) technologies provided new opportunities to better understand perfusion and microvasculature in the WM. Here, we aim to evaluate vasodilatory capacity of the WM vasculature, which is thought to be important in tissue ischemia and autoregulation. Fifteen younger and fifteen older subjects performed a CO2 inhalation task while blood-oxygenation-level-dependent (BOLD) magnetic resonance imaging (MRI) images were continuously collected. The cerebrovascular reactivity (CVR) index showed that the value of CVR in the WM (0.03±0.002%/mm Hg) was positive, but was significantly lower than that in the gray matter (GM) (0.22±0.01%/mm Hg). More strikingly, the WM response showed a temporal delay of 19±3 seconds compared with GM, which was attributed to the longer time it takes for extravascular CO2 to change. With age, WM CVR response becomes greater and faster, which is opposite to the changes seen in the GM. These data suggest that characteristics of WM CVR are different from that of GM and caution should be used when interpreting pathologic WM CVR results.

Measuring cerebrovascular reactivity: what stimulus to use?

Cerebrovascular reactivity is the change in cerebral blood flow in response to a vasodilatory or vasoconstrictive stimulus. Measuring variations of cerebrovascular reactivity between different regions of the brain has the potential to not only advance understanding of how the cerebral vasculature controls the distribution of blood flow but also to detect cerebrovascular pathophysiology. While there are standardized and repeatable methods for estimating the changes in cerebral blood flow in response to a vasoactive stimulus, the same cannot be said for the stimulus itself. Indeed, the wide variety of vasoactive challenges currently employed in these studies impedes comparisons between them. This review therefore critically examines the vasoactive stimuli in current use for their ability to provide a standard repeatable challenge and for the practicality of their implementation. Such challenges include induced reductions in systemic blood pressure, and the administration of vasoactive substances such as acetazolamide and carbon dioxide. We conclude that many of the stimuli in current use do not provide a standard stimulus comparable between individuals and in the same individual over time. We suggest that carbon dioxide is the most suitable vasoactive stimulus. We describe recently developed computer-controlled MRI compatible gas delivery systems which are capable of administering reliable and repeatable vasoactive CO2 stimuli.