fMRI Noise Research Articles

CORSICA: correction of structured noise in fMRI by automatic identification of ICA components

When applied to functional magnetic resonance imaging (fMRI) data, spatial independent component analysis (sICA), a data-driven technique that addresses the blind source separation problem, seems able to extract components specifically related to physiological noise and brain movements. These components should be removed from the data to achieve structured noise reduction and improve any subsequent detection and analysis of signal fluctuations related to neural activity. We propose a new automatic method called CORSICA (CORrection of Structured noise using spatial Independent Component Analysis) to identify the components related to physiological noise, using prior information on the spatial localization of the main physiological fluctuations in fMRI data. As opposed to existing spectral priors, which may be subject to aliasing effects for long-TR data sets (typically acquired with TR >1 s), such spatial priors can be applied to fMRI data, regardless of the TR of the acquisitions. By comparing the proposed automatic selection to a manual selection performed visually by a human operator, we first show that CORSICA is able to identify the noise-related components for long-TR data with a high sensitivity and a specificity of 1. On short-TR data sets, we validate that the proposed method of noise reduction allows a substantial improvement of the signal-to-noise ratio evaluated at the cardiac and respiratory frequencies, even in the gray matter, while preserving the main fluctuations related to neural activity.

Effect of spatial smoothing on physiological noise in high-resolution fMRI

Physiological noise dominates the SNR of the fMRI time-course at commonly used spatial resolutions at field strengths of 3 T and above. Operating in this physiological noise dominated regime limits some benefits of high field acquisition since increases in image SNR produce only modest increases in time-course SNR. Although previous studies have shown that the physiological noise dominance can be mitigated by using higher spatial resolutions, not all functional studies require voxel sizes smaller than the thickness of the human cortex. In this study, we examine the effect of acquiring high spatial resolution, thermal noise dominated time-courses and spatially smoothing the images to lower resolutions, which would otherwise be physiological noise dominated. At high field strengths, where physiological noise is most problematic, this strategy lowered the overall time-course variance compared to direct acquisition at commonly used spatial resolution. At 7 T for example, 5 × 5 × 3 mm 3 resolution images derived from smoothing 1.5 × 1.5 × 3 mm 3 data improved time-course SNR by a factor of 1.89 compared to a time-series acquired at 5 × 5 × 3 mm 3. Presumably, this effect was derived from the reduced physiological-to-thermal noise ratio in the high spatial resolution data followed by a smoothing operation that improves SNR without adding physiological noise. Our findings demonstrate that in contrast to conventional SNR penalties associated with spatially smoothing Fourier data, the time-course SNR of smoothed high-resolution data can be improved compared to direct acquisition at the desired resolution.

A penalized likelihood approach to magnetic resonance image reconstruction

Currently, images acquired via magnetic resonance imaging (MRI) and functional magnetic resonance imaging (fMRI) technology are reconstructed using the discrete inverse Fourier transform. While computationally convenient, this approach is not able to filter out noise. This is a serious limitation because the amount of noise in MRI and fMRI can be substantial. In this paper, we propose an alternative approach to reconstruction, based on penalized likelihood methodology. In particular, we focus on non-linear shrinkage estimators and show that this approach achieves a great reduction in integrated mean squared error (IMSE) of the estimated image with respect to the currently used estimator. This approach is extremely fast and easy to implement computationally. In addition, it can be combined with various alternative approaches to MR image reconstruction and can be easily adapted to other, non-MRI contexts, in which the observed data and the quantities of interest are related via a linear transform.

WAVELET-BASED ESTIMATION OF HEMODYNAMIC RESPONSE FUNCTION FROM fMRI DATA

We present a new algorithm to estimate hemodynamic response function (HRF) and drift components of fMRI data in wavelet domain. The HRF is modeled by both parametric and nonparametric models. The functional Magnetic resonance Image (fMRI) noise is modeled as a fractional brownian motion (fBm). The HRF parameters are estimated in wavelet domain by exploiting the property that wavelet transforms with a sufficient number of vanishing moments decorrelates a fBm process. Using this property, the noise covariance matrix in wavelet domain can be assumed to be diagonal whose entries are estimated using the sample variance estimator at each scale. We study the influence of the sampling rate of fMRI time series and shape assumption of HRF on the estimation performance. Results are presented by adding synthetic HRFs on simulated and null fMRI data. We also compare these methods with an existing method,(1) where correlated fMRI noise is modeled by a second order polynomial functions.

Frequency specific impairment of automatic pitch change detection by fMRI acoustic noise: An MEG study

The loud acoustic noise produced by the magnetic resonance scanner is a major source of interference in auditory fMRI research. Whole-head magnetoencephalography (MEG) was used to investigate the interaction between the frequency range of auditory stimulation and fMRI acoustic noise. Pure tones and 3-harmonic complexes varying between 240 and 1240 Hz in frequency were presented while participants attended to a silent subtitled film. Continuous fMRI acoustic noise was presented during half of the blocks. The activity in six regions of interest was analyzed in 100–200 and 200–300 ms time windows to evaluate the magnetic counterparts of the mismatch negativity (MMN) and P3a brain responses. The results suggested that fMRI noise significantly reduced the amplitude of these responses. The effect of the noise on the automatic processing of the tones was more prominent for the tones with frequencies higher than 500 Hz. It is recommended that in the MMN protocols using continuous fMRI acquisition the sound stimuli should be spectrally separated from the fMRI scanner noise spectrum.

Spatial and Temporal Characteristics of Physiological Noise in fMRI at 3T

Physiological noise in blood oxygen level-dependent functional magnetic resonance imaging (BOLD fMRI) has been shown to have characteristics similar to the BOLD signal itself, suggesting that it may have a vascular dependence. In this study, we evaluated the influence of physiological noise in fMRI as revealed by the differences in vasculature sensitivity of gradient-echo echo-planar imaging (GE-EPI) and spin-echo EPI (SE-EPI). The contribution of physiological noise to the fMRI signal during activation of the visual cortex was assessed by comparing its temporal characteristics with respect to echo time (TE), using both GE-EPI and SE-EPI. The correlation of the noise in fMRI with apparent diffusion coefficient (ADC) and the number of components required to describe its variance, as determined by principal-component analysis (PCA), were also assessed. The SE-EPI data were less affected by a TE-dependence of noise, in contrast to the apparent physiological noise in GE-EPI. Voxel-wise analysis revealed that total apparent noise increased as ADC values increased, and the relationship was different for GE-EPI and SE-EPI. PCA revealed that while the number of components characterizing the noise in SE-EPI data increased in a TE-dependent manner, approaching that of white noise at long echo time, the number of components from GE-EPI data was TE-independent. The difference in sensitivities to physiological noise between SE-EPI and GE-EPI suggests that extravascular BOLD processes around draining veins contribute significantly to physiological noise in BOLD fMRI, and the suppression of this noise component may enhance SE-EPI BOLD sensitivity at higher fields.

Nonstationary noise estimation in functional MRI

An important issue in functional MRI analysis is accurate characterisation of the noise processes present in the data. Whilst conventional fMRI noise representations often assume stationarity (or time-invariance) in the noise generating sources, such approaches may serve to suppress important dynamic information about brain function. As an alternative to these fixed temporal assumptions, we present in this paper two time-varying procedures for examining nonstationary noise structure in fMRI data. In the first procedure, we approximate nonstationary behaviour by means of a collection of simple but numerous time-varying parametric models. This is accomplished through the derivation of a locally parametric AutoRegressive (AR) plus drift model which tracks temporal covariance by allowing the model parameters to evolve over time. Before exploring time variation in these parameters, window-widths (bandwidths) that are well suited to the latent time-varying noise structure must be determined. To do this, we employ a bandwidth selection mechanism based on Stein's Unbiased Risk Estimator (SURE) criterion. In the second procedure, we describe the fMRI noise using a nonparametric method based on Functional Data Analysis (FDA). This process generates well-conditioned nonstationary covariance estimates that reflect temporal continuity in the underlying data structure whilst penalizing effective model dimension. We demonstrate both methods on simulated data and investigate the presence of nonstationary noise in resting fMRI data using the whitening capabilities of the locally parametric procedure. We evaluate the comparative behaviour of the stationary and nonstationary AR-based methods on data acquired at 1.5, 3 and 7 T magnetic field strengths and show that incorporation of time variation in the AR parameters leads to an overall decrease in the level of residual structure in the data. The FDA noise modelling technique is formulated within an activation mapping procedure and compared to the SPM (Statistical Parametric Mapping) toolbox on a cognitive face recognition task. Both the SPM and FDA methods show good sensitivity on this task, but we find that inclusion of the nonstationary FDA noise model seems to improve detection power in important task-related medial temporal regions.

Non-white noise in fMRI: Does modelling have an impact?

The sources of non-white noise in Blood Oxygenation Level Dependent (BOLD) functional magnetic resonance imaging (fMRI) are many. Familiar sources include low-frequency drift due to hardware imperfections, oscillatory noise due to respiration and cardiac pulsation and residual movement artefacts not accounted for by rigid body registration. These contributions give rise to temporal autocorrelation in the residuals of the fMRI signal and invalidate the statistical analysis as the errors are no longer independent. The low-frequency drift is often removed by high-pass filtering, and other effects are typically modelled as an autoregressive (AR) process. In this paper, we propose an alternative approach: Nuisance Variable Regression (NVR). By inclusion of confounding effects in a general linear model (GLM), we first confirm that the spatial distribution of the various fMRI noise sources is similar to what has already been described in the literature. Subsequently, we demonstrate, using diagnostic statistics, that removal of these contributions reduces first and higher order autocorrelation as well as non-normality in the residuals, thereby improving the validity of the drawn inferences. In addition, we also compare the performance of the NVR method to the whitening approach implemented in SPM2.

Study of temporal stationarity and spatial consistency of fMRI noise using independent component analysis

Spatial independent component analysis (ICA) was used to study the temporal stationarity and spatial consistency of structured functional MRI (fMRI) noise. Spatial correlations have been used in the past to generate filters for the removal of structured noise for each time-course in an fMRI dataset. It would be beneficial to produce a multivariate filter based on the same principles. ICA is examined to determine if it has properties that are beneficial for this type of filtering. Six fMRI baseline datasets were decomposed via spatial ICA. The time-courses associated with each component were tested for wide-sense stationarity using the wide sense stationarity quotient (WSS). Each dataset was divided into three subsets and each subset was decomposed. The components of first and third subset were matched by the strength of their correlation. The components produced by ICA were found to have largely nonstationary time-courses. Despite the temporal nonstationarity in the data, ICA was found to produce consistent spatial components. The degree of correlation among components differed depending on the amount of dimension reduction performed on the data. It was found that a relatively small number of dimensions produced components that are potentially useful for generating a spatial fMRI filter.

Extraction of overt verbal response from the acoustic noise in a functional magnetic resonance imaging scan by use of segmented active noise cancellation

A method to extract the subject's overt verbal response from the obscuring acoustic noise in an fMRI scan is developed by applying active noise cancellation with a conventional MRI microphone. Since the EPI scanning and its accompanying acoustic noise in fMRI are repetitive, the acoustic noise in one time segment was used as a reference noise in suppressing the acoustic noise in subsequent segments. However, the acoustic noise from the scanner was affected by the subject's movements, so the reference noise was adaptively adjusted as the scanner's acoustic properties varied in time. This method was successfully applied to a cognitive fMRI experiment with overt verbal responses.

Resampling fMRI time series

The problem of selecting a threshold for the statistical parameter maps in functional MRI (fMRI) is a delicate issue. The use of advanced test statistics and/or the complex dependence structure of fMRI noise may preclude parametric statistical methods for finding appropriate thresholds. Non-parametric statistical methodology has been presented as a feasible alternative. In this paper, we discuss resampling methods for finding thresholds in single subject fMRI analysis. It is shown that the presence of a BOLD response in the time series biases the estimation of the temporal autocorrelation, which in turn leads to biased thresholds. Therefore, proposed resampling methods based on Fourier and wavelet transforms, which employ implicit and weak models of the temporal noise characteristic, may produce erroneous thresholds. In contrast, resampling based on a pre-whitening transform, which is driven by an explicit noise model, is robust to the presence of a BOLD response. The size of the bias is, however, largely dependent on the complexity of the experimental design. While blocked designs can induce large biases, event-related designs generate significantly smaller biases. Results supporting these claims are provided.

Fractional Gaussian noise, functional MRI and Alzheimer's disease

Fractional Gaussian noise (fGn) provides a parsimonious model for stationary increments of a self-similar process parameterised by the Hurst exponent, H, and variance, σ 2. Fractional Gaussian noise with H < 0.5 demonstrates negatively autocorrelated or antipersistent behaviour; fGn with H > 0.5 demonstrates 1/ f, long memory or persistent behaviour; and the special case of fGn with H = 0.5 corresponds to classical Gaussian white noise. We comparatively evaluate four possible estimators of fGn parameters, one method implemented in the time domain and three in the wavelet domain. We show that a wavelet-based maximum likelihood (ML) estimator yields the most efficient estimates of H and σ 2 in simulated fGn with 0 < H < 1. Applying this estimator to fMRI data acquired in the “resting” state from healthy young and older volunteers, we show empirically that fGn provides an accommodating model for diverse species of fMRI noise, assuming adequate preprocessing to correct effects of head movement, and that voxels with H > 0.5 tend to be concentrated in cortex whereas voxels with H < 0.5 are more frequently located in ventricles and sulcal CSF. The wavelet-ML estimator can be generalised to estimate the parameter vector β for general linear modelling (GLM) of a physiological response to experimental stimulation and we demonstrate nominal type I error control in multiple testing of β, divided by its standard error, in simulated and biological data under the null hypothesis β = 0. We illustrate these methods principally by showing that there are significant differences between patients with early Alzheimer's disease (AD) and age-matched comparison subjects in the persistence of fGn in the medial and lateral temporal lobes, insula, dorsal cingulate/medial premotor cortex, and left pre- and postcentral gyrus: patients with AD had greater persistence of resting fMRI noise (larger H) in these regions. Comparable abnormalities in the AD patients were also identified by a permutation test of local differences in the first-order autoregression AR(1) coefficient, which was significantly more positive in patients. However, we found that the Hurst exponent provided a more sensitive metric than the AR(1) coefficient to detect these differences, perhaps because neurophysiological changes in early AD are naturally better described in terms of abnormal salience of long memory dynamics than a change in the strength of association between immediately consecutive time points. We conclude that parsimonious mapping of fMRI noise properties in terms of fGn parameters efficiently estimated in the wavelet domain is feasible and can enhance insight into the pathophysiology of Alzheimer's disease.

Evaluation of mixed effects in event-related fMRI studies: impact of first-level design and filtering

With the introduction of event-related designs in fMRI, it has become crucial to optimize design efficiency and temporal filtering to detect activations at the 1st level with high sensitivity. We investigate the relevance of these issues for fMRI population studies, that is, 2nd-level analysis, for a set of event-related fMRI (er-fMRI) designs with different 1st-level efficiencies, adopting three distinct 1st-level filtering strategies as implemented in SPM99, SPM2, and FSL3.0. By theory, experiments, and simulations using physiological fMRI noise, we show that both design and filtering impact the outcome of the statistical analysis, not only at the 1st but also at the 2nd level. There are several reasons behind this finding. First, sensitivity is affected by both design and filtering, since the scan-to-scan variance, that is the fixed effect, is not negligible with respect to the between-subject variance, that is the random effect, in er-fMRI population studies. The impact of the fixed effects error on the sensitivity of the mixed effects analysis can be mitigated by an optimal choice of er-fMRI design and filtering. Moreover, the accuracy of the 1st- and 2nd-level parameter estimates also depend on design and filtering; especially, we show that inaccuracies caused by the presence of residual noise autocorrelations can be constrained by designs that have hemodynamic responses with a Gaussian distribution. In conclusion, designs with both good efficiency and decorrelating properties, for example, such as the geometric or Latin square probability distributions, combined with the “whitening” filters of SPM2 and FSL3.0, give the best result, both for 1st- and 2nd-level analysis of er-fMRI studies.

Acoustic noise concerns in functional magnetic resonance imaging.

Magnetic resonance (MR) acoustic scanner noise may negatively affect the performance of functional magnetic resonance imaging (fMRI), a problem that worsens at the higher field strengths proposed to enhance fMRI. We present an overview of the current knowledge on the effects of confounding acoustic MR noise in fMRI experiments. The principles and effectiveness of various methods to reduce acoustic noise in fMRI are discussed, practical considerations are addressed and recommendations are made.

Effects of noise from functional magnetic resonance imaging on auditory event-related potentials in working memory task

The effects of functional magnetic resonance imaging (fMRI) acoustic noise were investigated on the parameters of event-related responses (ERPs) elicited during auditory matching-to-sample location and pitch working memory tasks. Stimuli were tones with varying location (left or right) and frequency (high or low). Subjects were instructed to memorize and compare either the locations or frequencies of the stimuli with each other. Tape-recorded fMRI acoustic noise was presented in half of the experimental blocks. The fMRI noise considerably enhanced the P1 component, reduced the amplitude and increased the latency of the N1, shortened the latency of the N2, and enhanced the amplitude of the P3 in both tasks. The N1 amplitude was higher in the location than pitch task in both noise and no-noise blocks, whereas the task-related N1 latency difference was present in the no-noise blocks only. Although the task-related differences between spatial and nonspatial auditory responses were partially preserved in noise, the finding that the acoustic gradient noise accompanying functional MR imaging modulated the auditory ERPs implies that the noise may confound the results of auditory fMRI experiments especially when studying higher cognitive processing.

Experimental model for functional magnetic resonance imaging of somatic sensory cortex in the unanesthetized rat

Functional magnetic resonance imaging (fMRI) has evolved into a method widely used to map neural activation in the human brain. fMRI is a method for recording blood oxygen level-dependent (BOLD) signals. These signals change with local cerebral blood flow coupled to neural activity. However, the relationship between BOLD signals and neural function is poorly understood and requires the development of animal models. Here we use an unanesthetized rat preparation to study BOLD responses to whisker stimulation in somatic sensory barrel cortex. Five rats were trained to tolerate restraint in a holder and fMRI noise with positive reinforcement. For maximal immobilization, the head was fastened to the holder with nuts screwed on threaded bolts attached to the head. On scanning day, residual stress was alleviated with injections of diazepam, and the rats were restrained in the holder and transferred into the scanner. After >75 min to allow the tranquilization to abate, structural images were acquired from three coronal brain slices. Subsequently, functional images were taken utilizing 4-min epochs without stimulation alternated with equivalent epochs during which the right caudal whiskers were stimulated with three air puffs/s. After 4 weeks, fMRI could be repeated in four rats. In seven of the nine functional runs, head motion was minimal and whisker stimulation resulted in a statistically significant ( P ≤ 0.05) increase in BOLD signal in barrel cortex predominantly on the contralateral side. The results provide encouragement that long-term fMRI studies on cerebral function in unanesthetized rats may be feasible with our procedure.

Efficient Design of Event-Related fMRI Experiments Using M-Sequences

Rapid event-related fMRI (erfMRI) allows estimation of the shape of hemodynamic responses (HDR) associated with transient brain activation evoked by various sensory, motor, and cognitive events. Choosing a sequence of events that maximizes efficiency of estimating the HDR is essential for conducting event-related brain imaging experiments, since increasing efficiency is essentially equivalent to reducing scanning time or increasing the strength of the principal magnetic field. The efficiency of an erfMRI design depends critically on the temporal arrangement of the sequence of events and the noise in the fMRI signal. We introduce to erfMRI a simple method for generating efficient event sequences based on maximum-length shift register sequences, or m-sequences. We show that under the assumption of white uncorrelated MRI noise, efficiency of erfMRI experimental designs that employ m-sequences exceeds efficiency of the best randomly generated sequences. This is true for single and multiple event type experiments, which allow either parallel events (overlapping events design) or designs in which only one event occurs at a time (nonoverlapping events design). HDR estimation efficiency afforded by m-sequences grows with the number of event types, and is greatest when event sequences are relatively short, albeit within commonly used scan times (i.e., 63–255 total events per scan). The improvement in efficiency, however, comes at a cost of constraints imposed by m-sequence generation rules, such as predetermined sequence lengths; for nonoverlapping events design m-sequence-based designs are not available for all possible numbers of event types. Nevertheless, designs that are available with m-sequences cover a large subset of commonly used erfMRI experimental designs. Under conditions of characteristic time-correlated fMRI noise, randomly generated sequences may yield efficiencies that exceed those afforded by m-sequences for single event-type designs, since in this case one can generate random sequences that partially decorrelate MRI noise by chance. Our simulations suggest that for designs of realistic sequence lengths that use more than one event type, m-sequence based designs tend to outperform random designs, thus making the knowledge of noise inessential. Finally, within an r-th order m-sequence (generated by a shift register of length r) all possible combinations of subsequences of length r occur, and thus these subsequences are exactly counterbalanced. This property is essential for minimizing effects of psychological and neuronal adaptation and expectation.

Analysis and use of FMRI response delays.

In this study, we implemented a new method for measuring the temporal delay of functional magnetic resonance imaging (fMRI) responses and then estimated the statistical distribution of response delays evoked by visual stimuli (checkered annuli) within and across voxels in human visual cortex. We assessed delay variability among different cortical sites and between parenchyma and blood vessels. Overall, 81% of all responsive voxels showed activation in phase with the stimulus while the remaining voxels showed antiphase, suppressive responses. Mean delays for activated and suppressed voxels were not significantly different (P < 0.001). Cortical flat maps showed that the pattern of activated and suppressed voxels was dynamically induced and depended on stimulus size. Mean delays for blood vessels were 0.7-2.4 sec longer than for parenchyma (P < 0.01). However, both parenchyma and blood vessels produced responses with long delays. We developed a model to identify and quantify different components contributing to variability in the empirical delay measurements. Within-voxel changes in delay over time were fully accounted for by the effects of empirically measured fMRI noise with virtually no measurable variability associated with the stimulus-induced response itself. Across voxels, as much as 47% of the delay variance was also the result of fMRI noise, with the remaining variance reflecting fixed differences in response delay among brain sites. In all cases, the contribution of fMRI noise to the delay variance depended on the noise power at the stimulus frequency. White noise models significantly underestimated the fMRI noise effects.

Isolating the auditory system from acoustic noise during functional magnetic resonance imaging: examination of noise conduction through the ear canal, head, and body.

Approaches were examined for reducing acoustic noise levels heard by subjects during functional magnetic resonance imaging (fMRI), a technique for localizing brain activation in humans. Specifically, it was examined whether a device for isolating the head and ear canal from sound (a "helmet") could add to the isolation provided by conventional hearing protection devices (i.e., earmuffs and earplugs). Both subjective attenuation (the difference in hearing threshold with versus without isolation devices in place) and objective attenuation (difference in ear-canal sound pressure) were measured. In the frequency range of the most intense fMRI noise (1-1.4 kHz), a helmet, earmuffs, and earplugs used together attenuated perceived sound by 55-63 dB, whereas the attenuation provided by the conventional devices alone was substantially less: 30-37 dB for earmuffs, 25-28 dB for earplugs, and 39-41 dB for earmuffs and earplugs used together. The data enabled the clarification of the relative importance of ear canal, head, and body conduction routes to the cochlea under different conditions: At low frequencies (< or =500 Hz), the ear canal was the dominant route of sound conduction to the cochlea for all of the device combinations considered. At higher frequencies (>500 Hz), the ear canal was the dominant route when either earmuffs or earplugs were worn. However, the dominant route of sound conduction was through the head when both earmuffs and earplugs were worn, through both ear canal and body when a helmet and earmuffs were worn, and through the body when a helmet, earmuffs, and earplugs were worn. It is estimated that a helmet, earmuffs, and earplugs together will reduce the most intense fMRI noise levels experienced by a subject to 60-65 dB SPL. Even greater reductions in noise should be achievable by isolating the body from the surrounding noise field.

Analysis of fMRI signal and noise component TE dependence

Analysis of fMRI signal and noise component TE dependence