Fractional Gaussian noise, functional MRI and Alzheimer's disease
Introduction
It is well-known that functional magnetic resonance imaging (fMRI) time series typically demonstrate complex and locally variable autocorrelation structure, even when the data have been acquired with the subject “at rest”. There is preliminary evidence that fMRI noise often has long memory in time, or 1/f spectral properties, meaning it is positively autocorrelated and there is disproportionate power in the spectrum at low frequencies (Zarahn et al., 1997); for review, see Bullmore et al. (2004) and references therein. However, there may also be high-frequency events like spikes or transients, or more sustained bursts of negative autocorrelation, in fMRI noise.
Not only is fMRI noise diverse, its presumably multiple sources are incompletely known, and are likely to vary in impact from one data set to the next. Head movement, for example, is an individually variable but common source of long memory noise caused by slow rotation or translation of the subject's head through an imperfectly homogeneous magnetic or RF field during scanning (Bullmore et al., 1999a). Cardiac and/or respiratory cycle-related pulsations may also contribute noise with properties depending in part on the sampling rate of data acquisition (repetition time, TR) and the proportion of cerebrospinal fluid represented in a voxel (Cordes et al., 2001, Purdon and Weisskoff, 1999). There are inevitably also instrumental and thermal sources of noise.
To date, one of the most successful modelling strategies for fMRI noise G = (G1,G2,…,Gn), n being the number of time points, has been the adoption of autoregressive, linear time invariant models (Bullmore et al., 1996, Bullmore et al., 2001, Dale, 1999, Locascio et al., 1997, Purdon and Weisskoff, 1999, Worsley et al., 2002) of the formwhere p is the order of the autoregressive AR(p) process and t = 1,2,…,n. However, AR models will require many parameters to account for long-range autocorrelated processes. The variability of autocorrelation between voxels (Bullmore et al., 1996, Worsley et al., 2002) suggests that it might be appropriate to adapt the order of AR process to each individual time series, which can be automated using model selection criteria such as the Bayesian information criterion (Fadili and Bullmore, 2002), but this is not always done in practice.
In this paper, we consider an alternative class of models, called fractional Gaussian noise or fGn, as a new approach to statistical modelling of fMRI noise. More formally, we are interested in the model that fMRI noise is distributed as a fractional Gaussian noise specified completely by only two parameters, its Hurst exponent H and its variance σ2, i.e., G ∼ N(0,Σ), with covariance matrix Σ ≔ Σ(H,σ2). The main potential advantage of fGn compared to AR(p) models is its simplicity and parsimony: Only two parameters need to be estimated, with no complications concerning optimal model order specification.
In the rest of this paper, we rehearse some key mathematical properties of fractional Gaussian noise; we comparatively evaluate four possible estimators of fGn parameters; and we apply the optimal, wavelet-based maximum likelihood estimator to analysis of fMRI noise properties in resting data acquired from healthy volunteers and patients with early Alzheimer's disease, motivated by the hypothesis that there might be disease-related changes in the Hurst exponent of fMRI noise (Jeong, 2004).
Section snippets
Terms and notation
A fractional Gaussian noise G = (Gt: t = 1,2,…, n) is a zero mean stationary process characterised by two parameters: the Hurst exponent H ∈ (0,1) and the variance σ2 ≔ Var (Gt), . The distribution of fractional Gaussian noise is fully specified by its auto-covariances at lags (Beran, 1994),
As can be seen from Eq. (2), the Hurst exponent is a measure of the long-term correlation between the discrete time points Gt, whereas the variance is only a scale
Comparative evaluation of fGn estimators
Using simulated fGn with known parameters, we comparatively evaluated the performance of Whittle's estimator in the time domain and three estimators in the wavelet domain: the wavelet-ML estimator incorporating the exact expression for the SDF; the estimator based on discrete variations of a filtered fractional Brownian motion; and Wornell's algorithm. The wavelet-ML algorithm was also used to estimate the parameter vector β of an arbitrary design matrix X fitted to simulated fGn and real fMRI
Discussion
The problem of fMRI noise is not new and there have been several previous efforts to address it statistically, many of them based on linear time-invariant, autoregressive and/or moving average models (Bullmore et al., 1996, Dale, 1999, Locascio et al., 1997, Purdon and Weisskoff, 1999, Worsley et al., 2002). Here we have introduced fractional Gaussian noise as an alternative modelling strategy. We have shown empirically that the spectral exponents of movement-corrected fMRI data acquired at
Acknowledgments
This neuroinformatics research was supported by a Human Brain Project grant from the National Institute of Biomedical Imaging and Bioengineering and the National Institute of Mental Health. fMRI data acquisition at 1.5 T was supported by a Wellcome Trust grant to RH and at 3.0 T was supported by an MRC cooperative group grant to the Wolfson Brain Imaging Centre (WBIC). We thank colleagues in the MRI Unit at the Maudsley Hospital, London, and the WBIC, Cambridge, for technical assistance with
References (63)
- et al.
Generic brain activation mapping in functional magnetic resonance imaging: a nonparametric approach
Magn. Reson. Imaging
(1997) - et al.
Topographic EEG changes with normal ageing and SDAT
Electroencephalogr. Clin. Neurophysiol.
(1989) - et al.
Wavelets and functional magnetic resonance imaging of the human brain
NeuroImage
(2004) Alzheimer's disease: more than a cholinergic disorder—evidence that cholinergic–monoaminergic interactions contribute to EEG slowing and dementia
Behav. Brain Res.
(2000)- et al.
Wavelet-generalised least squares: a new BLU estimator of linear regression model with 1/f errors
NeuroImage
(2002) - et al.
Quadratic variations and estimation of the Hölder index of a Gaussian process
Ann. Inst. Henri Poincaré
(1997) EEG dynamics in patients with Alzheimer's disease
Clin. Neurophysiol.
(2004)- et al.
EEG coherence in Alzheimer's disease
Electroencephalogr. Clin. Neurophysiol.
(1998) - et al.
Effects of a single intravenous does of scopolamine on the quantitative EEG in Alzheimer's disease patients and age-matched controls
Electroencephalogr. Clin. Neurophysiol.
(1994) - et al.
Wavelet-based multifractal analysis of fMRI time series
NeuroImage
(2004)