An Automated,Adaptive Framework for Optimizing Preprocessing Pipelines in Task-Based Functional MRI

BOLD fMRI is sensitive to blood-oxygenation changes correlated with brain function; however, it is limited by relatively weak signal and significant noise confounds. Many preprocessing algorithms have been developed to control noise and improve signal detection in fMRI. Although the chosen set of preprocessing and analysis steps (the “pipeline”) significantly affects signal detection, pipelines are rarely quantitatively validated in the neuroimaging literature, due to complex preprocessing interactions. This paper outlines and validates an adaptive resampling framework for evaluating and optimizing preprocessing choices by optimizing data-driven metrics of task prediction and spatial reproducibility. Compared to standard “fixed” preprocessing pipelines, this optimization approach significantly improves independent validation measures of within-subject test-retest, and between-subject activation overlap, and behavioural prediction accuracy. We demonstrate that preprocessing choices function as implicit model regularizers, and that improvements due to pipeline optimization generalize across a range of simple to complex experimental tasks and analysis models. Results are shown for brief scanning sessions (<3 minutes each), demonstrating that with pipeline optimization, it is possible to obtain reliable results and brain-behaviour correlations in relatively small datasets.     

Comparison between principal component analysis and independent component analysis in electroencephalograms modelling

Principal Component Analysis (PCA) is a classical technique in statistical data analysis, feature extraction and data reduction, aiming at explaining observed signals as a linear combination of orthogonal principal components. Independent Component Analysis (ICA) is a technique of array processing and data analysis, aiming at recovering unobserved signals or 'sources' from observed mixtures, exploiting only the assumption of mutual independence between the signals. The separation of the sources by ICA has great potential in applications such as the separation of sound signals (like voices mixed in simultaneous multiple records, for example), in telecommunication or in the treatment of medical signals. However, ICA is not yet often used by statisticians. In this paper, we shall present ICA in a statistical framework and compare this method with PCA for electroencephalograms (EEG) analysis.We shall see that ICA provides a more useful data representation than PCA, for instance, for the representation of a particular characteristic of the EEG named event-related potential (ERP).     

Temporally and Spatially Constrained ICA of fMRI Data Analysis

Constrained independent component analysis (CICA) is capable of eliminating the order ambiguity that is found in the standard ICA and extracting the desired independent components by incorporating prior information into the ICA contrast function. However, the current CICA method produces constraints that are based on only one type of prior information (temporal/spatial), which may increase the dependency of CICA on the accuracy of the prior information. To improve the robustness of CICA and to reduce the impact of the accuracy of prior information on CICA, we proposed a temporally and spatially constrained ICA (TSCICA) method that incorporated two types of prior information, both temporal and spatial, as constraints in the ICA. The proposed approach was tested using simulated fMRI data and was applied to a real fMRI experiment using 13 subjects who performed a movement task. Additionally, the performance of TSCICA was compared with the ICA method, the temporally CICA (TCICA) method and the spatially CICA (SCICA) method. The results from the simulation and from the real fMRI data demonstrated that TSCICA outperformed TCICA, SCICA and ICA in terms of robustness to noise. Moreover, the TSCICA method displayed better robustness to prior temporal/spatial information than the TCICA/SCICA method.     

Removal of Pulse Artefact from EEG Data Recorded in MR Environment at 3T. Setting of ICA Parameters for Marking Artefactual Components: Application to Resting-State Data

Simultaneous electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) allow for a non-invasive investigation of cerebral functions with high temporal and spatial resolution. The main challenge of such integration is the removal of the pulse artefact (PA) that affects EEG signals recorded in the magnetic resonance (MR) scanner. Often applied techniques for this purpose are Optimal Basis Set (OBS) and Independent Component Analysis (ICA). The combination of OBS and ICA is increasingly used, since it can potentially improve the correction performed by each technique separately. The present study is focused on the OBS-ICA combination and is aimed at providing the optimal ICA parameters for PA correction in resting-state EEG data, where the information of interest is not specified in latency and amplitude as in, for example, evoked potential. A comparison between two intervals for ICA calculation and four methods for marking artefactual components was performed. The performance of the methods was discussed in terms of their capability to 1) remove the artefact and 2) preserve the information of interest. The analysis included 12 subjects and two resting-state datasets for each of them. The results showed that none of the signal lengths for the ICA calculation was highly preferable to the other. Among the methods for the identification of PA-related components, the one based on the wavelets transform of each component emerged as the best compromise between the effectiveness in removing PA and the conservation of the physiological neuronal content.     

Stochastic ICA contrast maximisation using OJA's nonlinear PCA algorithm

Independent Component Analysis (ICA) is an important extension of linear Principal Component Analysis (PCA). PCA performs a data transformation to provide independence to second order, that is, decorrelation. ICA transforms data to provide approximate independence up to and beyond second order yielding transformed data with fully factorable probability densities. The linear ICA transformation has been applied to the classical statistical signal-processing problem of Blind Separation of Sources (BSS), that is, separating unknown original source signals from a mixture whose mode of mixing is undetermined. In this paper it is shown that Oja's Nonlinear PCA algorithm performs a general stochastic online adaptive ICA. This analysis is corroborated with three simulations. The first separates unknown mixtures of original natural images, which have sub-Gaussian densities, the second separates linear mixtures of natural speech whose densities are super-Gaussian. Finally unknown mixtures of original images, which have both sub- and super-Gaussian densities are separated.     

A simple and objective method for reproducible resting state network (RSN) detection in fMRI

Spatial Independent Component Analysis (ICA) decomposes the time by space functional MRI (fMRI) matrix into a set of 1-D basis time courses and their associated 3-D spatial maps that are optimized for mutual independence. When applied to resting state fMRI (rsfMRI), ICA produces several spatial independent components (ICs) that seem to have biological relevance - the so-called resting state networks (RSNs). The ICA problem is well posed when the true data generating process follows a linear mixture of ICs model in terms of the identifiability of the mixing matrix. However, the contrast function used for promoting mutual independence in ICA is dependent on the finite amount of observed data and is potentially non-convex with multiple local minima. Hence, each run of ICA could produce potentially different IC estimates even for the same data. One technique to deal with this run-to-run variability of ICA was proposed by [1] in their algorithm RAICAR which allows for the selection of only those ICs that have a high run-to-run reproducibility. We propose an enhancement to the original RAICAR algorithm that enables us to assign reproducibility p-values to each IC and allows for an objective assessment of both within subject and across subjects reproducibility. We call the resulting algorithm RAICAR-N (N stands for null hypothesis test), and we have applied it to publicly available human rsfMRI data (http://www.nitrc.org). Our reproducibility analyses indicated that many of the published RSNs in rsfMRI literature are highly reproducible. However, we found several other RSNs that are highly reproducible but not frequently listed in the literature.     

Natural Image Coding in V1: How Much Use Is Orientation Selectivity?

Orientation selectivity is the most striking feature of simple cell coding in V1 that has been shown to emerge from the reduction of higher-order correlations in natural images in a large variety of statistical image models. The most parsimonious one among these models is linear Independent Component Analysis (ICA), whereas second-order decorrelation transformations such as Principal Component Analysis (PCA) do not yield oriented filters. Because of this finding, it has been suggested that the emergence of orientation selectivity may be explained by higher-order redundancy reduction. To assess the tenability of this hypothesis, it is an important empirical question how much more redundancy can be removed with ICA in comparison to PCA or other second-order decorrelation methods. Although some previous studies have concluded that the amount of higher-order correlation in natural images is generally insignificant, other studies reported an extra gain for ICA of more than 100%. A consistent conclusion about the role of higher-order correlations in natural images can be reached only by the development of reliable quantitative evaluation methods. Here, we present a very careful and comprehensive analysis using three evaluation criteria related to redundancy reduction: In addition to the multi-information and the average log-loss, we compute complete rate–distortion curves for ICA in comparison with PCA. Without exception, we find that the advantage of the ICA filters is small. At the same time, we show that a simple spherically symmetric distribution with only two parameters can fit the data significantly better than the probabilistic model underlying ICA. This finding suggests that, although the amount of higher-order correlation in natural images can in fact be significant, the feature of orientation selectivity does not yield a large contribution to redundancy reduction within the linear filter bank models of V1 simple cells.     

ECG beat classification using PCA,LDA, ICA and Discrete Wavelet Transform

Electrocardiogram (ECG) is the P-QRS-T wave, representing the cardiac function. The information concealed in the ECG signal is useful in detecting the disease afflicting the heart. It is very difficult to identify the subtle changes in the ECG in time and frequency domains. The Discrete Wavelet Transform (DWT) can provide good time and frequency resolutions and is able to decipher the hidden complexities in the ECG. In this study, five types of beat classes of arrhythmia as recommended by Association for Advancement of Medical Instrumentation (AAMI) were analyzed namely: non-ectopic beats, supra-ventricular ectopic beats, ventricular ectopic beats, fusion betas and unclassifiable and paced beats. Three dimensionality reduction algorithms; Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA) and Independent Component Analysis (ICA) were independently applied on DWT sub bands for dimensionality reduction. These dimensionality reduced features were fed to the Support Vector Machine (SVM), neural network (NN) and probabilistic neural network (PNN) classifiers for automated diagnosis. ICA features in combination with PNN with spread value (σ) of 0.03 performed better than the PCA and LDA. It has yielded an average sensitivity, specificity, positive predictive value (PPV) and accuracy of 99.97%, 99.83%, 99.21% and 99.28% respectively using ten-fold cross validation scheme.     

The correction of eye blink artefacts in the EEG: a comparison of two prominent methods

Background

The study investigated the residual impact of eyeblinks on the electroencephalogram (EEG) after application of different correction procedures, namely a regression method (eye movement correction procedure, EMCP) and a component based method (Independent Component Analysis, ICA).

Methodology/Principle Findings

Real and simulated data were investigated with respect to blink-related potentials and the residual mutual information of uncorrected vertical electrooculogram (EOG) and corrected EEG, which is a measure of residual EOG contribution to the EEG. The results reveal an occipital positivity that peaks at about 250ms after the maximum blink excursion following application of either correction procedure. This positivity was not observable in the simulated data. Mutual information of vertical EOG and EEG depended on the applied regression procedure. In addition, different correction results were obtained for real and simulated data. ICA yielded almost perfect correction in all conditions. However, under certain conditions EMCP yielded comparable results to the ICA approach.

Conclusion

In conclusion, for EMCP the quality of correction depended on the EMCP variant used and the structure of the data, whereas ICA always yielded almost perfect correction. However, its disadvantage is the much more complex data processing, and that it requires a suitable amount of data.     

Stimulus-related independent component and voxel-wise analysis of human brain activity during free viewing of a feature film

Understanding how the brain processes stimuli in a rich natural environment is a fundamental goal of neuroscience. Here, we showed a feature film to 10 healthy volunteers during functional magnetic resonance imaging (fMRI) of hemodynamic brain activity. We then annotated auditory and visual features of the motion picture to inform analysis of the hemodynamic data. The annotations were fitted to both voxel-wise data and brain network time courses extracted by independent component analysis (ICA). Auditory annotations correlated with two independent components (IC) disclosing two functional networks, one responding to variety of auditory stimulation and another responding preferentially to speech but parts of the network also responding to non-verbal communication. Visual feature annotations correlated with four ICs delineating visual areas according to their sensitivity to different visual stimulus features. In comparison, a separate voxel-wise general linear model based analysis disclosed brain areas preferentially responding to sound energy, speech, music, visual contrast edges, body motion and hand motion which largely overlapped the results revealed by ICA. Differences between the results of IC- and voxel-based analyses demonstrate that thorough analysis of voxel time courses is important for understanding the activity of specific sub-areas of the functional networks, while ICA is a valuable tool for revealing novel information about functional connectivity which need not be explained by the predefined model. Our results encourage the use of naturalistic stimuli and tasks in cognitive neuroimaging to study how the brain processes stimuli in rich natural environments.     

Default Mode,Dorsal Attention and Auditory Resting State Networks Exhibit Differential Functional Connectivity in Tinnitus and Hearing Loss

We investigated auditory, dorsal attention, and default mode networks in adults with tinnitus and hearing loss in a resting state functional connectivity study. Data were obtained using continuous functional magnetic resonance imaging (fMRI) while the participants were at “rest” and were not performing any task. Participants belonged to one of three groups: middle-aged adults with tinnitus and mild-to-moderate high frequency hearing loss (TIN), age-matched controls with normal hearing and no tinnitus (NH), and a second control group with mild-to-moderate high frequency hearing loss without tinnitus (HL). After standard preprocessing, (a) a group independent component analysis (ICA) using 30 components and (b) a seeding-based connectivity analysis were conducted. In the group ICA, the default mode network was the only network to display visual differences between subject groups. In the seeding analysis, we found increased connectivity between the left parahippocampus and the auditory resting state network in the TIN group when compared to NH controls. Similarly, there was also an increased correlation between the right parahippocampus and the dorsal attention network when compared to HL controls. Other group differences in this attention network included decreased correlations between the seed regions and the right supramarginal gyrus in TIN patients when compared to HL controls. In the default mode network, there was a strong decrease in correlation between the seed regions and the precuneus when compared to both control groups. The findings of this study identify specific alterations in the connectivity of the default mode, dorsal attention, and auditory resting state networks due to tinnitus. The results suggest that therapies for tinnitus that mitigate the increased connectivity of limbic regions with auditory and attention resting state networks and the decreased coherence of the default mode network could be effective at reducing tinnitus-related distress.     

A general probabilistic model for group independent component analysis and its estimation methods

Independent component analysis (ICA) has become an important tool for analyzing data from functional magnetic resonance imaging (fMRI) studies. ICA has been successfully applied to single-subject fMRI data. The extension of ICA to group inferences in neuroimaging studies, however, is challenging due to the unavailability of a prespecified group design matrix and the uncertainty in between-subjects variability in fMRI data. We present a general probabilistic ICA (PICA) model that can accommodate varying group structures of multisubject spatiotemporal processes. An advantage of the proposed model is that it can flexibly model various types of group structures in different underlying neural source signals and under different experimental conditions in fMRI studies. A maximum likelihood (ML) method is used for estimating this general group ICA model. We propose two expectation-maximization (EM) algorithms to obtain the ML estimates. The first method is an exact EM algorithm, which provides an exact E-step and an explicit noniterative M-step. The second method is a variational approximation EM algorithm, which is computationally more efficient than the exact EM. In simulation studies, we first compare the performance of the proposed general group PICA model and the existing probabilistic group ICA approach. We then compare the two proposed EM algorithms and show the variational approximation EM achieves comparable accuracy to the exact EM with significantly less computation time. An fMRI data example is used to illustrate application of the proposed methods.     

Separate, causal roles of the caudate in saccadic choice and execution in a perceptual decision task

In contrast to the well-established roles of the striatum in movement generation and value-based decisions, its contributions to perceptual decisions lack direct experimental support. Here, we show that electrical microstimulation in the monkey caudate nucleus influences both choice and saccade response time on a visual motion discrimination task. Within a drift-diffusion framework, these effects consist of two components. The perceptual component biases choices toward ipsilateral targets, away from the neurons' predominantly contralateral response fields. The choice bias is consistent with a nonzero starting value of the diffusion process, which increases and decreases decision times for contralateral and ipsilateral choices, respectively. The nonperceptual component decreases and increases nondecision times toward contralateral and ipsilateral targets, respectively, consistent with the caudate's role in saccade generation. The results imply a causal role for the caudate in perceptual decisions used to select saccades that may be distinct from its role in executing those saccades. VIDEO ABSTRACT:     

Motion edges and regions guide image segmentation by colour.

Image segmentation is an important early stage in visual processing in which the visual system groups together parts of the image that belong together, prior to or in conjunction with object recognition. Two principal processes may be involved in image segmentation: an edge-based process that uses feature contrasts to mark boundaries of coherent regions, and a region-based process that groups similar features over a larger scale. Earlier, we have shown that motion and colour interact strongly in image segmentation by the human visual system. Here we explore the nature of this interaction in terms of edge- and region-based processes. We measure performance on a region-based colour segmentation task in the presence of distinct types of motion information, in the form of edges and regions which in themselves do not reveal the location of the colour target. The results show that both motion edges and regions may guide the integrative process required for this colour segmentation task. Motion edges appear to act by delimiting areas over which to integrate colour information, whereas motion similarities define primitive surfaces within which colour grouping and segmentation processes are deployed.     

Robust Data Driven Model Order Estimation for Independent Component Analysis of fMRI Data with Low Contrast to Noise

Independent component analysis (ICA) has been successfully utilized for analysis of functional MRI (fMRI) data for task related as well as resting state studies. Although it holds the promise of becoming an unbiased data-driven analysis technique, a few choices have to be made prior to performing ICA, selection of a method for determining the number of independent components (nIC) being one of them. Choice of nIC has been shown to influence the ICA maps, and various approaches (mostly relying on information theoretic criteria) have been proposed and implemented in commonly used ICA analysis packages, such as MELODIC and GIFT. However, there has been no consensus on the optimal method for nIC selection, and many studies utilize arbitrarily chosen values for nIC. Accurate and reliable determination of true nIC is especially important in the setting where the signals of interest contribute only a small fraction of the total variance, i.e. very low contrast-to-noise ratio (CNR), and/or very focal response. In this study, we evaluate the performance of different model order selection criteria and demonstrate that the model order selected based upon bootstrap stability of principal components yields more reliable and accurate estimates of model order. We then demonstrate the utility of this fully data-driven approach to detect weak and focal stimulus-driven responses in real data. Finally, we compare the performance of different multi-run ICA approaches using pseudo-real data.     

Contributions of the amygdala to reward expectancy and choice signals in human prefrontal cortex

The prefrontal cortex (PFC) receives substantial anatomical input from the amygdala, and these two structures have long been implicated in reward-related learning and decision making. Yet little is known about how these regions interact, especially in humans. We investigated the contribution of the amygdala to reward-related signals in PFC by scanning two rare subjects with focal bilateral amygdala lesions using fMRI. The subjects performed a reversal learning task in which they first had to learn which of two choices was the more rewarding, and then flexibly switch their choices when contingencies changed. Compared with healthy controls, both amygdala lesion subjects showed a profound change in ventromedial prefrontal cortex (vmPFC) activity associated with reward expectation and behavioral choice. These findings support a critical role for the human amygdala in establishing expected reward representations in PFC, which in turn may be used to guide behavioral choice.     

Bayesian independent component analysis recovers pathway signatures from blood metabolomics data

Interpreting the complex interplay of metabolites in heterogeneous biosamples still poses a challenging task. In this study, we propose independent component analysis (ICA) as a multivariate analysis tool for the interpretation of large-scale metabolomics data. In particular, we employ a Bayesian ICA method based on a mean-field approach, which allows us to statistically infer the number of independent components to be reconstructed. The advantage of ICA over correlation-based methods like principal component analysis (PCA) is the utilization of higher order statistical dependencies, which not only yield additional information but also allow a more meaningful representation of the data with fewer components. We performed the described ICA approach on a large-scale metabolomics data set of human serum samples, comprising a total of 1764 study probands with 218 measured metabolites. Inspecting the source matrix of statistically independent metabolite profiles using a weighted enrichment algorithm, we observe strong enrichment of specific metabolic pathways in all components. This includes signatures from amino acid metabolism, energy-related processes, carbohydrate metabolism, and lipid metabolism. Our results imply that the human blood metabolome is composed of a distinct set of overlaying, statistically independent signals. ICA furthermore produces a mixing matrix, describing the strength of each independent component for each of the study probands. Correlating these values with plasma high-density lipoprotein (HDL) levels, we establish a novel association between HDL plasma levels and the branched-chain amino acid pathway. We conclude that the Bayesian ICA methodology has the power and flexibility to replace many of the nowadays common PCA and clustering-based analyses common in the research field.     

Binocular interaction: contrast matching and contrast discrimination are predicted by the same model

How do signals from the 2 eyes combine and interact? Our recent work has challenged earlier schemes in which monocular contrast signals are subject to square-law transduction followed by summation across eyes and binocular gain control. Much more successful was a new 'two-stage' model in which the initial transducer was almost linear and contrast gain control occurred both pre- and post-binocular summation. Here we extend that work by: (i) exploring the two-dimensional stimulus space (defined by left- and right-eye contrasts) more thoroughly, and (ii) performing contrast discrimination and contrast matching tasks for the same stimuli. Twenty-five base-stimuli made from 1 c/deg patches of horizontal grating, were defined by the factorial combination of 5 contrasts for the left eye (0.3-32%) with five contrasts for the right eye (0.3-32%). Other than in contrast, the gratings in the two eyes were identical. In a 2IFC discrimination task, the base-stimuli were masks (pedestals), where the contrast increment was presented to one eye only. In a matching task, the base-stimuli were standards to which observers matched the contrast of either a monocular or binocular test grating. In the model, discrimination depends on the local gradient of the observer's internal contrast-response function, while matching equates the magnitude (rather than gradient) of response to the test and standard. With all model parameters fixed by previous work, the two-stage model successfully predicted both the discrimination and the matching data and was much more successful than linear or quadratic binocular summation models. These results show that performance measures and perception (contrast discrimination and contrast matching) can be understood in the same theoretical framework for binocular contrast vision.