Methods to detect, characterize, and remove motion artifact in resting state fMRI.
ABSTRACT: Head motion systematically alters correlations in resting state functional connectivity fMRI (RSFC). In this report we examine impact of motion on signal intensity and RSFC correlations. We find that motion-induced signal changes (1) are often complex and variable waveforms, (2) are often shared across nearly all brain voxels, and (3) often persist more than 10s after motion ceases. These signal changes, both during and after motion, increase observed RSFC correlations in a distance-dependent manner. Motion-related signal changes are not removed by a variety of motion-based regressors, but are effectively reduced by global signal regression. We link several measures of data quality to motion, changes in signal intensity, and changes in RSFC correlations. We demonstrate that improvements in data quality measures during processing may represent cosmetic improvements rather than true correction of the data. We demonstrate a within-subject, censoring-based artifact removal strategy based on volume censoring that reduces group differences due to motion to chance levels. We note conditions under which group-level regressions do and do not correct motion-related effects.
Project description:Recent fMRI studies have outlined the critical impact of in-scanner head motion, particularly on estimates of functional connectivity. Common strategies to reduce the influence of motion include realignment as well as the inclusion of nuisance regressors, such as the 6 realignment parameters, their first derivatives, time-shifted versions of the realignment parameters, and the squared parameters. However, these regressors have limited success at noise reduction. We hypothesized that using nuisance regressors consisting of the principal components (PCs) of edge voxel time series would be better able to capture slice-specific and nonlinear signal changes, thus explaining more variance, improving data quality (i.e., lower DVARS and temporal SNR), and reducing the effect of motion on default-mode network connectivity. Functional MRI data from 22 healthy adult subjects were preprocessed using typical motion regression approaches as well as nuisance regression derived from edge voxel time courses. Results were evaluated in the presence and absence of both global signal regression and motion censoring. Nuisance regressors derived from signal intensity time courses at the edge of the brain significantly improved motion correction compared to using only the realignment parameters and their derivatives. Of the models tested, only the edge voxel regression models were able to eliminate significant differences in default-mode network connectivity between high- and low-motion subjects regardless of the use of global signal regression or censoring.
Project description:Head motion is a significant source of noise in the estimation of functional connectivity from resting-state functional MRI (rs-fMRI). Current strategies to reduce this noise include image realignment, censoring time points corrupted by motion, and including motion realignment parameters and their derivatives as additional nuisance regressors in the general linear model. However, this nuisance regression approach assumes that the motion-induced signal changes are linearly related to the estimated realignment parameters, which is not always the case. In this study we develop an improved model of motion-related signal changes, where nuisance regressors are formed by first rotating and translating a single brain volume according to the estimated motion, re-registering the data, and then performing a principal components analysis (PCA) on the resultant time series of both moved and re-registered data. We show that these "Motion Simulated (MotSim)" regressors account for significantly greater fraction of variance, result in higher temporal signal-to-noise, and lead to functional connectivity estimates that are less affected by motion compared to the most common current approach of using the realignment parameters and their derivatives as nuisance regressors. This improvement should lead to more accurate estimates of functional connectivity, particularly in populations where motion is prevalent, such as patients and young children.
Project description:Subject motion degrades the quality of task functional magnetic resonance imaging (fMRI) data. Here, we test two classes of methods to counteract the effects of motion in task fMRI data: (1) a variety of motion regressions and (2) motion censoring ("motion scrubbing"). In motion regression, various regressors based on realignment estimates were included as nuisance regressors in general linear model (GLM) estimation. In motion censoring, volumes in which head motion exceeded a threshold were withheld from GLM estimation. The effects of each method were explored in several task fMRI data sets and compared using indicators of data quality and signal-to-noise ratio. Motion censoring decreased variance in parameter estimates within- and across-subjects, reduced residual error in GLM estimation, and increased the magnitude of statistical effects. Motion censoring performed better than all forms of motion regression and also performed well across a variety of parameter spaces, in GLMs with assumed or unassumed response shapes. We conclude that motion censoring improves the quality of task fMRI data and can be a valuable processing step in studies involving populations with even mild amounts of head movement.
Project description: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.
Project description:Resting-state functional magnetic resonance imaging (rs-fMRI) based on the blood-oxygen-level-dependent (BOLD) signal has been widely used in healthy individuals and patients to investigate brain functions when the subjects are in a resting or task-negative state. Head motion considerably confounds the interpretation of rs-fMRI data. Nuisance regression is commonly used to reduce motion-related artifacts with six motion parameters estimated from rigid-body realignment as regressors. To further compensate for the effect of head movement, the first-order temporal derivatives of motion parameters and squared motion parameters were proposed previously as possible motion regressors. However, these additional regressors may not be sufficient to model the impact of head motion because of the complexity of motion artifacts. In addition, while using more motion-related regressors could explain more variance in the data, the neural signal may also be removed with increasing number of motion regressors. To better model how in-scanner motion affects rs-fMRI data, a robust and automated convolutional neural network (CNN) model is developed in this study to obtain optimal motion regressors. The CNN network consists of two temporal convolutional layers and the output from the network are the derived motion regressors used in the following nuisance regression. The temporal convolutional layer in the network can non-parametrically model the prolonged effect of head motion. The set of regressors derived from the neural network is compared with the same number of regressors used in a traditional nuisance regression approach. It is demonstrated that the CNN-derived regressors can more effectively reduce motion-related artifacts.
Project description:Artifactual sources of resting-state (RS) FMRI can originate from head motion, physiology, and hardware. Of these sources, motion has received considerable attention and was found to induce corrupting effects by differentially biasing correlations between regions depending on their distance. Numerous corrective approaches have relied on the identification and censoring of high-motion time points and the use of the brain-wide average time series as a nuisance regressor to which the data are orthogonalized (Global Signal Regression, GSReg). We first replicate the previously reported head-motion bias on correlation coefficients using data generously contributed by Power et al. (2012). We then show that while motion can be the source of artifact in correlations, the distance-dependent bias-taken to be a manifestation of the motion effect on correlation-is exacerbated by the use of GSReg. Put differently, correlation estimates obtained after GSReg are more susceptible to the presence of motion and by extension to the levels of censoring. More generally, the effect of motion on correlation estimates depends on the preprocessing steps leading to the correlation estimate, with certain approaches performing markedly worse than others. For this purpose, we consider various models for RS FMRI preprocessing and show that WMeLOCAL, as subset of the ANATICOR discussed by Jo et al. (2010), denoising approach results in minimal sensitivity to motion and reduces by extension the dependence of correlation results on censoring.
Project description:Differing noise variance across study populations has been shown to cause artifactual group differences in functional connectivity measures. In this study, we investigate the use of short echo time functional MRI data to correct for these noise sources in blood oxygenation level dependent (BOLD)-weighted time series. A dual-echo sequence was used to simultaneously acquire data at both a short (TE=3.3 ms) and a BOLD-weighted (TE=35 ms) echo time. This approach is effectively "free," using dead-time in the pulse sequence to collect an additional echo without affecting overall scan time or temporal resolution. The proposed correction method uses voxelwise regression of the short TE data from the BOLD-weighted data to remove noise variance. In addition to a typical resting state scan, non-compliant behavior associated with patient groups was simulated via increased head motion or physiological fluctuations in 10 subjects. Short TE data showed significant correlations with the traditional motion-related and physiological noise regressors used in current connectivity analyses. Following traditional preprocessing, the extent of significant additional variance explained by the short TE data regressors was significantly correlated with the average head motion across the scan in the resting data (r(2)=0.93, p<0.0001). The reduction in data variance following the inclusion of short TE regressors was also correlated with scan head motion (r(2)=0.48, p=0.027). Task-related data were used to demonstrate the effects of the short TE correction on BOLD activation time series with known temporal structure; the size and strength of the activation were significantly decreased, but it is not clear whether this reflects BOLD contamination in the short TE data or correlated changes in blood volume. Finally, functional connectivity maps of the default mode network were constructed using a seed correlation approach. The effects of short TE correction and low-pass filtering on the resulting correlations maps were compared. Results suggest that short TE correction more accurately differentiates artifactual correlations from the correlations of interest in conditions of amplified noise.
Project description:In the last decade, an increasing interest has arisen in investigating the relationship between the electrophysiological and hemodynamic measurements of brain activity, such as EEG and (BOLD) fMRI. In particular, changes in BOLD have been shown to be associated with changes in the spectral profile of neural activity, rather than with absolute power. Concurrently, recent findings showed that different EEG rhythms are independently related to changes in the BOLD signal: therefore, it would be also important to distinguish between the contributions of the different EEG rhythms to BOLD fluctuations when modeling the relationship between the two signals. Here we propose a method to perform EEG-informed fMRI analysis where the changes in the spectral profile are modeled, and, at the same time, the distinction between rhythms is preserved. We compared our model with two other frequency-dependent regressors modeling using simultaneous EEG-fMRI data from healthy subjects performing a motor task. Our results showed that the proposed method better captures the correlations between BOLD signal and EEG rhythms modulations, identifying task-related, well localized activated volumes. Furthermore, we showed that including among the regressors also EEG rhythms not primarily involved in the task enhances the performance of the analysis, even when only correlations with BOLD signal and specific EEG rhythms are explored.
Project description:The simultaneous acquisition and subsequent analysis of EEG and fMRI data is challenging owing to increased noise levels in the EEG data. A common method to integrate data from these two modalities is to use aspects of the EEG data, such as the amplitudes of event-related potentials (ERP) or oscillatory EEG activity, to predict fluctuations in the fMRI data. However, this relies on the acquisition of high quality datasets to ensure that only the correlates of neuronal activity are being studied. In this study, we investigate the effects of head-motion-related artefacts in the EEG signal on the predicted T2*-weighted signal variation. We apply our analyses to two independent datasets: 1) four participants were asked to move their feet in the scanner to generate small head movements, and 2) four participants performed an episodic memory task. We created T2*-weighted signal predictors from indicators of abrupt head motion using derivatives of the realignment parameters, from visually detected artefacts in the EEG as well as from three EEG frequency bands (theta, alpha and beta). In both datasets, we found little correlation between the T2*-weighted signal and EEG predictors that were not convolved with the canonical haemodynamic response function (cHRF). However, all convolved EEG predictors strongly correlated with the T2*-weighted signal variation in various regions including the bilateral superior temporal cortex, supplementary motor area, medial parietal cortex and cerebellum. The finding that movement onset spikes in the EEG predict T2*-weighted signal intensity only when the time course of movements is convolved with the cHRF, suggests that the correlated signal might reflect a BOLD response to neural activity associated with head movement. Furthermore, the observation that broad-spectral EEG spikes tend to occur at the same time as abrupt head movements, together with the finding that abrupt movements and EEG spikes show similar correlations with the T2*-weighted signal, indicates that the EEG spikes are produced by abrupt movement and that continuous regressors of EEG oscillations contain motion-related noise even after stringent correction of the EEG data. If not properly removed, these artefacts complicate the use of EEG data as a predictor of T2*-weighted signal variation.
Project description:Global signal regression (GSR) is one of the most debated preprocessing strategies for resting-state functional MRI. GSR effectively removes global artifacts driven by motion and respiration, but also discards globally distributed neural information and introduces negative correlations between certain brain regions. The vast majority of previous studies have focused on the effectiveness of GSR in removing imaging artifacts, as well as its potential biases. Given the growing interest in functional connectivity fingerprinting, here we considered the utilitarian question of whether GSR strengthens or weakens associations between resting-state functional connectivity (RSFC) and multiple behavioral measures across cognition, personality and emotion. By applying the variance component model to the Brain Genomics Superstruct Project (GSP), we found that behavioral variance explained by whole-brain RSFC increased by an average of 47% across 23 behavioral measures after GSR. In the Human Connectome Project (HCP), we found that behavioral variance explained by whole-brain RSFC increased by an average of 40% across 58 behavioral measures, when GSR was applied after ICA-FIX de-noising. To ensure generalizability, we repeated our analyses using kernel regression. GSR improved behavioral prediction accuracies by an average of 64% and 12% in the GSP and HCP datasets respectively. Importantly, the results were consistent across methods. A behavioral measure with greater RSFC-explained variance (using the variance component model) also exhibited greater prediction accuracy (using kernel regression). A behavioral measure with greater improvement in behavioral variance explained after GSR (using the variance component model) also enjoyed greater improvement in prediction accuracy after GSR (using kernel regression). Furthermore, GSR appeared to benefit task performance measures more than self-reported measures. Since GSR was more effective at removing motion-related and respiratory-related artifacts, GSR-related increases in variance explained and prediction accuracies were unlikely the result of motion-related or respiratory-related artifacts. However, it is worth emphasizing that the current study focused on whole-brain RSFC, so it remains unclear whether GSR improves RSFC-behavioral associations for specific connections or networks. Overall, our results suggest that at least in the case for young healthy adults, GSR strengthens the associations between RSFC and most (although not all) behavioral measures. Code for the variance component model and ridge regression can be found here: https://github.com/ThomasYeoLab/CBIG/tree/master/stable_projects/preprocessing/Li2019_GSR.