Detection of consistently task-related activations in fMRI data with hybrid independent component analysis

Mapping dynamic spatial patterns of brain function with spatial-wise attention

Objective: Using functional magnetic resonance imaging (fMRI) and deep learning to discover the spatial pattern of brain function, or functional brain networks (FBNs) has been attracted many reseachers. Most existing works focus on static FBNs or dynamic functional connectivity among fixed spatial network nodes, but ignore the potential dynamic/time-varying characteristics of the spatial networks themselves. And most of works based on the assumption of linearity and independence, that oversimplify the relationship between blood-oxygen level dependence signal changes and the heterogeneity of neuronal activity within voxels.Approach: To overcome these problems, we proposed a novel spatial-wise attention (SA) based method called Spatial and Channel-wise Attention Autoencoder (SCAAE) to discover the dynamic FBNs without the assumptions of linearity or independence. The core idea of SCAAE is to apply the SA to generate FBNs directly, relying solely on the spatial information present in fMRI volumes. Specifically, we trained the SCAAE in a self-supervised manner, using the autoencoder to guide the SA to focus on the activation regions. Experimental results show that the SA can generate multiple meaningful FBNs at each fMRI time point, which spatial similarity are close to the FBNs derived by known classical methods, such as independent component analysis.Main results: To validate the generalization of the method, we evaluate the approach on HCP-rest, HCP-task and ADHD-200 dataset. The results demonstrate that SA mechanism can be used to discover time-varying FBNs, and the identified dynamic FBNs over time clearly show the process of time-varying spatial patterns fading in and out.Significance: Thus we provide a novel method to understand human brain better. Code is available athttps://github.com/WhatAboutMyStar/SCAAE.

Spatial-temporal convolutional attention for discovering and characterizing functional brain networks in task fMRI

Functional brain networks (FBNs) are spatial patterns of brain function that play a critical role in understanding human brain function. There are many proposed methods for mapping the spatial patterns of brain function, however they oversimplify the underlying assumptions of brain function and have various limitations such as linearity and independence. Additionally, current methods fail to account for the dynamic nature of FBNs, which limits their effectiveness in accurately characterizing these networks. To address these limitations, we present a novel deep learning and spatial-wise attention based model called Spatial-Temporal Convolutional Attention (STCA) to accurately model dynamic FBNs. Specifically, we train STCA in a self-supervised manner by utilizing a Convolutional Autoencoder to guide the STCA module in assigning higher attention weights to regions of functional activity. To validate the reliability of the results, we evaluate our approach on the HCP-task motor behavior dataset, the experimental results demonstrate that the STCA derived FBNs have higher spatial similarity with the templates and that the spatial similarity between the templates and the FBNs derived by STCA fluctuates with the task design over time, suggesting that STCA can reflect the dynamic changes of brain function, providing a powerful tool to better understand human brain function. Code is available at https://github.com/SNNUBIAI/STCAE.

Cross Domain Correlation Maximization for Enhancing the Target Recognition of SSVEP-Based Brain-Computer Interfaces

The target recognition performance of steady-state visual evoked potential (SSVEP)-based brain-computer interfaces can be significantly improved with a training-based approach. However, the training procedure is time consuming and often causes fatigue. Consequently, the number of training data should be limited, which may reduce the classification performance. Thus, how to improve classification accuracy without increasing the training time is crucial to SSVEP-based BCI system. This study proposes a transfer-related component analysis (TransRCA) method for addressing the above issue. In this method, the SSVEP-related components are extracted from a small number of training data of the current individual and combined with those extracted from a large number of existing training data of other individuals. The TransRCA method maximizes not only the inter-trial covariances between the source and target subjects, but also the correlation between the reference signals and SSVEP signals from the source and target subjects. The proposed method was validated on the SSVEP public Benchmark and BETA datasets, and the classification accuracy and information transmission rate of the ensemble version of the proposed TransRCA method were compared with those of the state-of-the-art eCCA, eTRCA, ttCCA, LSTeTRCA, and eIISMC methods on both datasets. The comparison results indicate that the proposed method provides a superior performance compared with these state-of-the-art methods, and thus has high potential for the development of a SSVEP-based brain-computer interface system with high classification performance that only uses a small number of training data.

A deep learning method for autism spectrum disorder identification based on interactions of hierarchical brain networks

BACKGROUND

It has been recently shown that deep learning models exhibited remarkable performance of representing functional Magnetic Resonance Imaging (fMRI) data for the understanding of brain functional activities. With hierarchical structure, deep learning models can infer hierarchical functional brain networks (FBN) from fMRI. However, the applications of the hierarchical FBNs have been rarely studied.

METHODS

In this work, we proposed a hierarchical recurrent variational auto-encoder (HRVAE) to unsupervisedly model the fMRI data. The trained HRVAE encoder can predict hierarchical temporal features from its three hidden layers, and thus can be regarded as a hierarchical feature extractor. Then LASSO (least absolute shrinkage and selection operator) regression was applied to estimate the corresponding hierarchical FBNs. Based on the hierarchical FBNs from each subject, we constructed a novel classification framework for brain disorder identification and test it on the Autism Brain Imaging Data Exchange (ABIDE) dataset, a world-wide multi-site database of autism spectrum disorder (ASD). We analyzed the hierarchy organization of FBNs, and finally used the overlaps of hierarchical FBNs as features to differentiate ASD from typically developing controls (TDC).

RESULTS

The experimental results on 871 subjects from ABIDE dataset showed that the HRVAE model can effectively derive hierarchical FBNs including many well-known resting state networks (RSN). Moreover, the classification result improved the state-of-the-art by achieving a very high accuracy of 82.1 %.

CONCLUSIONS

This work presents a novel data-driven deep learning method using fMRI data for ASD identification, which could provide valuable reference for clinical diagnosis. The classification results suggest that the interactions of hierarchical FBNs have association with brain disorder, which promotes the understanding of FBN hierarchy and could be applied to other brain disorder analysis.

Multi-head attention-based masked sequence model for mapping functional brain networks

The investigation of functional brain networks (FBNs) using task-based functional magnetic resonance imaging (tfMRI) has gained significant attention in the field of neuroimaging. Despite the availability of several methods for constructing FBNs, including traditional methods like GLM and deep learning methods such as spatiotemporal self-attention mechanism (STAAE), these methods have design and training limitations. Specifically, they do not consider the intrinsic characteristics of fMRI data, such as the possibility that the same signal value at different time points could represent different brain states and meanings. Furthermore, they overlook prior knowledge, such as task designs, during training. This study aims to overcome these limitations and develop a more efficient model by drawing inspiration from techniques in the field of natural language processing (NLP). The proposed model, called the Multi-head Attention-based Masked Sequence Model (MAMSM), uses a multi-headed attention mechanism and mask training approach to learn different states corresponding to the same voxel values. Additionally, it combines cosine similarity and task design curves to construct a novel loss function. The MAMSM was applied to seven task state datasets from the Human Connectome Project (HCP) tfMRI dataset. Experimental results showed that the features acquired by the MAMSM model exhibit a Pearson correlation coefficient with the task design curves above 0.95 on average. Moreover, the model can extract more meaningful networks beyond the known task-related brain networks. The experimental results demonstrated that MAMSM has great potential in advancing the understanding of functional brain networks.

Functional and effective connectivity analysis of drug-resistant epilepsy: a resting-state fMRI analysis

Objective

Epilepsy is considered as a neural network disorder. Seizure activity in epilepsy may disturb brain networks and damage brain functions. We propose using resting-state functional magnetic resonance imaging (rs-fMRI) data to characterize connectivity patterns in drug-resistant epilepsy.

Methods

This study enrolled 47 participants, including 28 with drug-resistant epilepsy and 19 healthy controls. Functional and effective connectivity was employed to assess drug-resistant epilepsy patients within resting state networks. The resting state functional connectivity (FC) analysis was performed to assess connectivity between each patient and healthy controls within the default mode network (DMN) and the dorsal attention network (DAN). In addition, dynamic causal modeling was used to compute effective connectivity (EC). Finally, a statistical analysis was performed to evaluate our findings.

Results

The FC analysis revealed significant connectivity changes in patients giving 64.3% (18/28) and 78.6% (22/28) for DMN and DAN, respectively. Statistical analysis of FC was significant between the medial prefrontal cortex, posterior cingulate cortex, and bilateral inferior parietal cortex for DMN. For DAN, it was significant between the left and the right intraparietal sulcus and the frontal eye field. For the DMN, the patient group showed significant EC connectivity in the right inferior parietal cortex and the medial prefrontal cortex for the DMN. There was also bilateral connectivity between the medial prefrontal cortex and the posterior cingulate cortex, as well as between the left and right inferior parietal cortex. For DAN, patients showed significant connectivity in the right frontal eye field and the right intraparietal sulcus. Bilateral connectivity was also found between the left frontal eye field and the left intraparietal sulcus, as well as between the right frontal eye field and the right intraparietal sulcus. The statistical analysis of the EC revealed a significant result in the medial prefrontal cortex and the right intraparietal cortex for the DMN. The DAN was found significant in the left frontal eye field, as well as the left and right intraparietal sulcus.

Conclusion

Our results provide preliminary evidence to support that the combination of functional and effective connectivity analysis of rs-fMRI can aid in diagnosing epilepsy in the DMN and DAN networks.

Learning brain representation using recurrent Wasserstein generative adversarial net

BACKGROUND AND OBJECTIVE

To understand brain cognition and disorders, modeling the mapping between mind and brain has been of great interest to the neuroscience community. The key is the brain representation, including functional brain networks (FBN) and their corresponding temporal features. Recently, it has been proven that deep learning models have superb representation power on functional magnetic resonance imaging (fMRI) over traditional machine learning methods. However, due to the lack of high-quality data and labels, deep learning models tend to suffer from overfitting in the training process.

METHODS

In this work, we applied a recurrent Wasserstein generative adversarial net (RWGAN) to learn brain representation from volumetric fMRI data. Generative adversarial net (GAN) is widely used in natural image generation and is able to capture the distribution of the input data, which enables the extraction of generalized features from fMRI and thus relieves the overfitting issue. The recurrent layers in RWGAN are designed to better model the local temporal features of the fMRI time series. The discriminator of RWGAN works as a deep feature extractor. With LASSO regression, the RWGAN model can decompose the fMRI data into temporal features and spatial features (FBNs). Furthermore, the generator of RWGAN can generate high-quality new data for fMRI augmentation.

RESULTS

The experimental results on seven tasks from the HCP dataset showed that the RWGAN can learn meaningful and interpretable temporal features and FBNs, compared to HCP task designs and general linear model (GLM) derived networks. Besides, the results on different training datasets showed that the RWGAN performed better on small datasets than other deep learning models. Moreover, we used the generator of RWGAN to yield fake subjects. The result showed that the fake data can also be used to learn meaningful representation compared to those learned from real data.

CONCLUSIONS

To our best knowledge, this work is among the earliest attempts of applying generative deep learning for modeling fMRI data. The proposed RWGAN offers a novel methodology for learning brain representation from fMRI, and it can generate high-quality fake data for the potential use of fMRI data augmentation.

Seed-based dual regression: An illustration of the impact of dual regression's inherent filtering of global signal

BACKGROUND

Functional connectivity (FC) maps from brain fMRI data are often derived with seed-based methods that estimate temporal correlations between the time course in a predefined region (seed) and other brain regions (SCA, seed-based correlation analysis). Standard dual regression, which uses a set of spatial regressor maps, can detect FC with entire brain "networks," such as the default mode network, but may not be feasible when detecting FC associated with a single small brain region alone (for example, the amygdala).

NEW METHOD

We explored seed-based dual regression (SDR) from theoretical and practical points of view. SDR is a modified implementation of dual regression where the set of spatial regressors is replaced by a single binary spatial map of the seed region.

RESULTS

SDR allowed detection of FC with small brain regions.

COMPARISON WITH EXISTING METHOD

For both synthetic and natural fMRI data, detection of FC with SDR was identical to that obtained with SCA after removal of global signal from fMRI data with global signal regression (GSR). In the absence of GSR, detection of FC was significantly improved when using SDR compared with SCA.

CONCLUSION

The improved FC detection achieved with SDR was related to a partial filtering of the global signal that occurred during spatial regression, an integral part of dual regression. This filtering can sometimes lead to spurious negative correlations that result in a widespread negative bias in FC derived with any application of dual regression. We provide guidelines for how to identify and correct this potential problem.

Modeling and augmenting of fMRI data using deep recurrent variational auto-encoder

Objective. Recently, deep learning models have been successfully applied in functional magnetic resonance imaging (fMRI) modeling and associated applications. However, there still exist at least two challenges. Firstly, due to the lack of sufficient data, deep learning models tend to suffer from overfitting in the training process. Secondly, it is still challenging to model the temporal dynamics from fMRI, due to that the brain state is continuously changing over scan time. In addition, existing methods rarely studied and applied fMRI data augmentation.Approach. In this work, we construct a deep recurrent variational auto-encoder (DRVAE) that combined variational auto-encoder and recurrent neural network, aiming to address all of the above mentioned challenges. The encoder of DRVAE can extract more generalized temporal features from assumed Gaussian distribution of input data, and the decoder of DRVAE can generate new data to increase training samples and thus partially relieve the overfitting issue. The recurrent layers in DRVAE are designed to effectively model the temporal dynamics of functional brain activities. LASSO (least absolute shrinkage and selection operator) regression is applied on the temporal features and input fMRI data to estimate the corresponding spatial networks.Main results. Extensive experimental results on seven tasks from HCP dataset showed that the DRVAE and LASSO framework can learn meaningful temporal patterns and spatial networks from both real data and generated data. The results on group-wise data and single subject suggest that the brain activities may follow certain distribution. Moreover, we applied DRVAE on four resting state fMRI datasets from ADHD-200 for data augmentation, and the results showed that the classification performances on augmented datasets have been considerably improved.Significance. The proposed method can not only derive meaningful temporal features and spatial networks from fMRI, but also generate high-quality new data for fMRI data augmentation and associated applications.

An Overview of ICA/BSS-Based Application to Alzheimer's Brain Signal Processing

Alzheimer's disease (AD) is by far the most common cause of dementia associated with aging. Early and accurate diagnosis of AD and ability to track progression of the disease is increasingly important as potential disease-modifying therapies move through clinical trials. With the advent of biomedical techniques, such as computerized tomography (CT), electroencephalography (EEG), magnetoencephalography (MEG), positron emission tomography (PET), magnetic resonance imaging (MRI), and functional magnetic resonance imaging (fMRI), large amounts of data from Alzheimer's patients have been acquired and processed from which AD-related information or "signals" can be assessed for AD diagnosis. It remains unknown how best to mine complex information from these brain signals to aid in early diagnosis of AD. An increasingly popular technique for processing brain signals is independent component analysis or blind source separation (ICA/BSS) that separates blindly observed signals into original signals that are as independent as possible. This overview focuses on ICA/BSS-based applications to AD brain signal processing.

Modeling task-based fMRI data via deep belief network with neural architecture search

It has been shown that deep neural networks are powerful and flexible models that can be applied on fMRI data with superb representation ability over traditional methods. However, a challenge of neural network architecture design has also attracted attention: due to the high dimension of fMRI volume images, the manual process of network model design is very time-consuming and not optimal. To tackle this problem, we proposed an unsupervised neural architecture search (NAS) framework on a deep belief network (DBN) that models volumetric fMRI data, named NAS-DBN. The NAS-DBN framework is based on Particle Swarm Optimization (PSO) where the swarms of neural architectures can evolve and converge to a feasible optimal solution. The experiments showed that the proposed NAS-DBN framework can quickly find a robust architecture of DBN, yielding a hierarchy organization of functional brain networks (FBNs) and temporal responses. Compared with 3 manually designed DBNs, the proposed NAS-DBN has the lowest testing loss of 0.0197, suggesting an overall performance improvement of up to 47.9 %. For each task, the NAS-DBN identified 260 FBNs, including task-specific FBNs and resting state networks (RSN), which have high overlap rates to general linear model (GLM) derived templates and independent component analysis (ICA) derived RSN templates. The average overlap rate of NAS-DBN to GLM on 20 task-specific FBNs is as high as 0.536, indicating a performance improvement of up to 63.9 % in respect of network modeling. Besides, we showed that the NAS-DBN can simultaneously generate temporal responses that resemble the task designs very well, and it was observed that widespread overlaps between FBNs from different layers of NAS-DBN model form a hierarchical organization of FBNs. Our NAS-DBN framework contributes an effective, unsupervised NAS method for modeling volumetric task fMRI data.

Omission of temporal nuisance regressors from dual regression can improve accuracy of fMRI functional connectivity maps

Functional connectivity (FC) maps from brain fMRI data can be derived with dual regression, a proposed alternative to traditional seed-based FC (SFC) methods that detect temporal correlation between a predefined region (seed) and other regions in the brain. As with SFC, incorporating nuisance regressors (NR) into the dual regression must be done carefully, to prevent potential bias and insensitivity of FC estimates. Here, we explore the potentially untoward effects on dual regression that may occur when NR correlate highly with the signal of interest, using both synthetic and real fMRI data to elucidate mechanisms responsible for loss of accuracy in FC maps. Our tests suggest significantly improved accuracy in FC maps derived with dual regression when highly correlated temporal NR were omitted. Single-map dual regression, a simplified form of dual regression that uses neither spatial nor temporal NR, offers a viable alternative whose FC maps may be more easily interpreted, and in some cases be more accurate than those derived with standard dual regression.

The potential of MR-Encephalography for BCI/Neurofeedback applications with high temporal resolution

Real-time functional magnetic resonance imaging (rt-fMRI) enables the update of various brain-activity measures during an ongoing experiment as soon as a new brain volume is acquired. However, the recorded Blood-oxygen-level dependent (BOLD) signal also contains physiological artifacts such as breathing and heartbeat, which potentially cause misleading false positive effects especially problematic in brain-computer interface (BCI) and neurofeedback (NF) setups. The low temporal resolution of echo planar imaging (EPI) sequences (which is in the range of seconds) prevents a proper separation of these artifacts from the BOLD signal. MR-Encephalography (MREG) has been shown to provide the high temporal resolution required to unalias and correct for physiological fluctuations and leads to increased specificity and sensitivity for mapping task-based activation and functional connectivity as well as for detecting dynamic changes in connectivity over time. By comparing a simultaneous multislice echo planar imaging (SMS-EPI) sequence and an MREG sequence using the same nominal spatial resolution in an offline analysis for three different experimental fMRI paradigms (perception of house and face stimuli, motor imagery, Stroop task), the potential of this novel technique for future BCI and NF applications was investigated. First, adapted general linear model pre-whitening which accounts for the high temporal resolution in MREG was implemented to calculate proper statistical results and be able to compare these with the SMS-EPI sequence. Furthermore, the respiration- and cardiac pulsation-related signals were successfully separated from the MREG signal using independent component analysis which were then included as regressors for a GLM analysis. Only the MREG sequence allowed to clearly separate cardiac pulsation and respiration components from the signal time course. It could be shown that these components highly correlate with the recorded respiration and cardiac pulsation signals using a respiratory belt and fingertip pulse plethysmograph. Temporal signal-to-noise ratios of SMS-EPI and MREG were comparable. Functional connectivity analysis using partial correlation showed a reduced standard error in MREG compared to SMS-EPI. Also, direct time course comparisons by down-sampling the MREG signal to the SMS-EPI temporal resolution showed lower variance in MREG. In general, we show that the higher temporal resolution is beneficial for fMRI time course modeling and this aspect can be exploited in offline application but also, is especially attractive, for real-time BCI and NF applications.

Task-Evoked Negative BOLD Response in the Default Mode Network Does Not Alter Its Functional Connectivity

While functional connectivity networks are often extracted from resting-state fMRI scans, they have been shown to be active during task performance as well. However, the effect of an in-scanner task on functional connectivity networks is not completely understood. While there is evidence that task-evoked positive BOLD response can alter functional connectivity networks, particularly in the primary sensorimotor cortices, the effect of task-evoked negative BOLD response on the functional connectivity of the Default mode network (DMN) is somewhat ambiguous. In this study, we aim to investigate whether task performance, which is associated with negative BOLD response in the DMN regions, alters the time-course of functional connectivity in the same regions obtained by independent component analysis (ICA). ICA has been used to effectively extract functional connectivity networks during task performance and resting-state. We first demonstrate that performing a simple visual-motor task alters the temporal time-course of the network extracted from the primary visual cortex. Then we show that despite detecting a robust task-evoked negative BOLD response in the DMN regions, a simple visual-motor task does not alter the functional connectivity of the DMN regions. Our findings suggest that different mechanisms may underlie the relationship between task-related activation/deactivation networks and the overlapping functional connectivity networks in the human large-scale brain networks.

Ten Key Observations on the Analysis of Resting-state Functional MR Imaging Data Using Independent Component Analysis

For more than 20 years, the powerful, flexible family of independent component analysis (ICA) techniques has been used to examine spatial, temporal, and subject variation in functional magnetic resonance (fMR) imaging data. This article provides an overview of 10 key principles in the basic and advanced application of ICA to resting-state fMR imaging. ICA's core advantages include robustness to artifact; false-positives and autocorrelation; adaptability to variant study designs; agnosticism to the temporal evolution of fMR imaging signals; and ability to extract, identify, and analyze neural networks. ICA remains in the vanguard of fMRI methods development.

Reduction of Motion Artifacts and Noise Using Independent Component Analysis in Task-Based Functional MRI for Preoperative Planning in Patients with Brain Tumor

BACKGROUND AND PURPOSE

Although it is a potentially powerful presurgical tool, fMRI can be fraught with artifacts, leading to interpretive errors, many of which are not fully accounted for in routinely applied correction methods. The purpose of this investigation was to evaluate the effects of data denoising by independent component analysis in patients undergoing preoperative evaluation for glioma resection compared with more routinely applied correction methods such as realignment or motion scrubbing.

MATERIALS AND METHODS

Thirty-five functional runs (both motor and language) in 12 consecutive patients with glioma were analyzed retrospectively by double-blind review. Data were processed and compared with the following: 1) realignment alone, 2) motion scrubbing, 3) independent component analysis denoising, and 4) both independent component analysis denoising and motion scrubbing. Primary outcome measures included a change in false-positives, false-negatives, z score, and diagnostic rating.

RESULTS

Independent component analysis denoising reduced false-positives in 63% of studies versus realignment alone. There was also an increase in the z score in areas of true activation in 71.4% of studies. Areas of new expected activation (previous false-negatives) were revealed in 34.4% of cases with independent component analysis denoising versus motion scrubbing or realignment alone. Of studies deemed nondiagnostic with realignment or motion scrubbing alone, 65% were considered diagnostic after independent component analysis denoising.

CONCLUSIONS

The addition of independent component analysis denoising of fMRI data in preoperative patients with glioma has a significant impact on data quality, resulting in reduced false-positives and an increase in true-positives compared with more commonly applied motion scrubbing or simple realignment methods.

Rethinking the Use of Neutral Faces as a Baseline in fMRI Neuroimaging Studies of Axis-I Psychiatric Disorders

Major Axis-I disorders including major depressive disorder (MDD), bipolar disorder (BD), anxiety disorder, and schizophrenia are associated with a host of aberrations in the way social stimuli are processed. Face perception tasks are often used in neuroimaging research of emotion processing in both healthy and patient populations, and to date, there exists a mounting body of evidence, both behavioral and within the brain, indicating that emotional faces compared to neutral faces are processed abnormally by those with Axis-I disorders relative to healthy control (HC) groups. The use of neutral faces as a "baseline control condition" is predicated on the assumption that neutral faces are processed in the same way HCs and individuals with major Axis-I disorders. In this paper, existing fMRI studies examining the way neutral faces are processed in groups with Axis-I disorders involving socioaffective perception are reviewed. In reviewing available studies, a consistent pattern of results demonstrated that these disorders are associated with abnormal frontolimbic activity in response to neutral faces and in particular within the amygdala and prefrontal regions such as the dorsolateral prefrontal cortex (DLPFC) and anterior cingulate cortex (ACC) compared to HC groups. Specifically, increased amygdala activation was consistently reported in response to neutral faces in anxiety disorders and schizophrenia. Abnormal medial PFC activity was reported in patients with MDD, and patients with BD exhibit decreased activity in the DLPFC and ACC relative to HCs. In addition, specific suggestions to overcome these obstacles with new research and additional analyses are discussed.

Predictors and brain connectivity changes associated with arm motor function improvement from intensive practice in chronic stroke

Background and Purpose: The brain changes that underlie therapy-induced improvement in motor function after stroke remain obscure. This study sought to demonstrate the feasibility and utility of measuring motor system physiology in a clinical trial of intensive upper extremity rehabilitation in chronic stroke-related hemiparesis. Methods: This was a substudy of two multi-center clinical trials of intensive robotic and intensive conventional therapy arm therapy in chronic, significantly hemiparetic, stroke patients. Transcranial magnetic stimulation was used to measure motor cortical output to the biceps and extensor digitorum communus muscles. Magnetic resonance imaging (MRI) was used to determine the cortical anatomy, as well as to measure fractional anisotropy, and blood oxygenation (BOLD) during an eyes-closed rest state. Region-of-interest time-series correlation analysis was performed on the BOLD signal to determine interregional connectivity. Functional status was measured with the upper extremity Fugl-Meyer and Wolf Motor Function Test. Results: Motor evoked potential (MEP) presence was associated with better functional outcomes, but the effect was not significant when considering baseline impairment. Affected side internal capsule fractional anisotropy was associated with better function at baseline. Affected side primary motor cortex (M1) activity became more correlated with other frontal motor regions after treatment. Resting state connectivity between affected hemisphere M1 and dorsal premotor area (PMAd) predicted recovery. Conclusions: Presence of motor evoked potentials in the affected motor cortex and its functional connectivity with PMAd may be useful in predicting recovery. Functional connectivity in the motor network shows a trends towards increasing after intensive robotic or non-robotic arm therapy. Clinical Trial Registration URL: http://www.clinicaltrials.gov. Unique identifiers: NCT00372411 \& NCT00333983.

Brain activation differences in schizophrenia during context-dependent processing of saccade tasks

Background

Brain function in schizophrenia has been probed using saccade paradigms and functional magnetic resonance imaging, but little information exists about how changing task context impacts saccade related brain activation and behavioral performance. We recruited schizophrenia and comparison subjects to perform saccade tasks in differing contexts: (1) two single task runs (anti- or pro-saccades alternating with fixation) and (2) one dual task run (antisaccades alternating with prosaccades).

Results

Context-dependent differences in saccade circuitry were evaluated using ROI analyses. Distinction between anti- and pro-saccade activation across contexts (single versus dual task) suggests that the schizophrenia group did not respond to context in the same way as the comparison group.

Conclusions

Further investigation of context processing effects on brain activation and saccade performance measures informs models of cognitive deficits in the disorder and enhances understanding of antisaccades as a potential endophenotype for schizophrenia.

Electronic supplementary material

The online version of this article (doi:10.1186/s12993-016-0103-2) contains supplementary material, which is available to authorized users.

The absence of task-related increases in BOLD signal does not equate to absence of task-related brain activation

Most fMRI studies employ general-linear-model-based analyses (GLM-BA) of BOLD signal changes to identify regions that are active (or not) during specific cognitive processes. However, alternate analytic approaches (like independent component analysis) may identify more complex patterns of activation, including in regions not implicated in GLM-BA of the same data. In our opinion, fMRI findings revealed by a GLM-BA cannot exclude any brain regions from contributing to specific cognitive processes.

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.

Review of fMRI Data Analysis

Classification of brain states obtained through functional magnetic resonance imaging (fMRI) poses a serious challenges for neuroimaging community to uncover discriminating patterns of brain state activity that define independent thought processes. This challenge came into existence because of the large number of voxels in a typical fMRI scan, the classifier is presented with a massive feature set coupled with a relatively small training samples. One of the most popular research topics in last few years is the application of machine learning algorithms for mental states classification, decoding brain activation, and finding the variable of interest from fMRI data. In classification scenario, different algorithms have different biases, in the sequel performances differs across datasets, and for a particular dataset the accuracy varies from classifier to classifier. To overcome the limitations of individual techniques, hybridization or fusion of these machine learning techniques emerged in recent years which have shown promising result and open up new direction of research. This paper reviews the machine learning techniques ranging from individual classifiers, ensemble, and hybrid techniques used in cognitive classification with a well balance treatment of their applications, performance, and limitations. It also discusses many open research challenges for further research.

Spatiotemporal independent component analysis combine general linear model applied to fMRI for eliminating neural noise

Functional magnetic resonance imaging (fMRI) has recently become an effective means to explore the mechanism of functional rehabilitation in stroke patients. Neural noise is an inevitable structural noise, and is an important factor caused individual differences in fMRI data, therefore, eliminating the neural noise is being regarded as one of the task that cannot be ignored. In this paper, a new algorithm combines spatiotemporal independent component analysis and general linear model (GLM) is proposed to eliminate the effect caused by excess neural activity. This new algorithm simultaneously maximizes the independence over time and space in fMRI data for establishing the spatiotemporal balance. The new technique was applied to extract the active regions of ankle dorsiflexion during fMRI scanning process. Compared to results of GLM, the results of new combined algorithm is more reasonable with an 8% improvement in correlation coefficient. It confirmed that this new algorithm is effective in eliminating system noise and neural disturbance.

Manual selection of spontaneous activity maps derived from independent component analysis: criteria and inter-rater reliability study

BACKGROUND

During the last years, many investigations focused on spontaneously active cerebral networks such as the default-mode network. A data-driven technique, the independent component analysis, allows segregating such spontaneous (co-)activity maps (SAM) from noise in functional magnetic resonance imaging (fMRI) time series. The inter-rater reliability of manual selection of not only the default-mode network but all SAMs remained to be assessed.

NEW METHOD

The current study was performed on 20 min (400 volumes) fMRI time series of 30 healthy participants. SAMs' selection criteria were first established on past experience and from the literature. The inter-rater reliability of SAMs vs non-SAMs manual selection was then investigated from 250 independent components per participant.

RESULTS

Inter-rater Kappa coefficient was of 0.89 ± 0.01 on whole analysis, and 0.88 ± 0.09 on participant per participant analysis.

COMPARISON WITH EXISTING METHODS

Without focusing on specific and predetermined SAMs only, our criteria allow a reliable selection of all SAMs including the idiosyncratic networks.

CONCLUSIONS

The proposed SAM's selection criteria are reliable enough to allow scientific exploration of all SAMs at the single subject level.

Applying independent component analysis to clinical FMRI at 7 t

Increased BOLD sensitivity at 7 T offers the possibility to increase the reliability of fMRI, but ultra-high field is also associated with an increase in artifacts related to head motion, Nyquist ghosting, and parallel imaging reconstruction errors. In this study, the ability of independent component analysis (ICA) to separate activation from these artifacts was assessed in a 7 T study of neurological patients performing chin and hand motor tasks. ICA was able to isolate primary motor activation with negligible contamination by motion effects. The results of General Linear Model (GLM) analysis of these data were, in contrast, heavily contaminated by motion. Secondary motor areas, basal ganglia, and thalamus involvement were apparent in ICA results, but there was low capability to isolate activation in the same brain regions in the GLM analysis, indicating that ICA was more sensitive as well as more specific. A method was developed to simplify the assessment of the large number of independent components. Task-related activation components could be automatically identified via these intuitive and effective features. These findings demonstrate that ICA is a practical and sensitive analysis approach in high field fMRI studies, particularly where motion is evoked. Promising applications of ICA in clinical fMRI include presurgical planning and the study of pathologies affecting subcortical brain areas.

Spatiotemporal Segregation of Neural Response to Auditory Stimulation: An fMRI Study Using Independent Component Analysis and Frequency-Domain Analysis

Although auditory processing has been widely studied with conventional parametric methods, there have been a limited number of independent component analysis (ICA) applications in this area. The purpose of this study was to examine spatiotemporal behavior of brain networks in response to passive auditory stimulation using ICA. Continuous broadband noise was presented binaurally to 19 subjects with normal hearing. ICA was performed to segregate spatial networks, which were subsequently classified according to their temporal relation to the stimulus using power spectrum analysis. Classification of separated networks resulted in 3 stimulus-activated, 9 stimulus-deactivated, 2 stimulus-neutral (stimulus-dependent but not correlated with the stimulation timing), and 2 stimulus-unrelated (fluctuations that did not follow the stimulus cycles) components. As a result of such classification, spatiotemporal subdivisions were observed in a number of cortical structures, namely auditory, cingulate, and sensorimotor cortices, where parts of the same cortical network responded to the stimulus with different temporal patterns. The majority of the classified networks seemed to comprise subparts of the known resting-state networks (RSNs); however, they displayed different temporal behavior in response to the auditory stimulus, indicating stimulus-dependent temporal segregation of RSNs. Only one of nine deactivated networks coincided with the “classic” default-mode network, suggesting the existence of a stimulus-dependent default-mode network, different from that commonly accepted.

Time course based artifact identification for independent components of resting-state FMRI

In functional magnetic resonance imaging (fMRI) coherent oscillations of the blood oxygen level-dependent (BOLD) signal can be detected. These arise when brain regions respond to external stimuli or are activated by tasks. The same networks have been characterized during wakeful rest when functional connectivity of the human brain is organized in generic resting-state networks (RSN). Alterations of RSN emerge as neurobiological markers of pathological conditions such as altered mental state. In single-subject fMRI data the coherent components can be identified by blind source separation of the pre-processed BOLD data using spatial independent component analysis (ICA) and related approaches. The resulting maps may represent physiological RSNs or may be due to various artifacts. In this methodological study, we propose a conceptually simple and fully automatic time course based filtering procedure to detect obvious artifacts in the ICA output for resting-state fMRI. The filter is trained on six and tested on 29 healthy subjects, yielding mean filter accuracy, sensitivity and specificity of 0.80, 0.82, and 0.75 in out-of-sample tests. To estimate the impact of clearly artifactual single-subject components on group resting-state studies we analyze unfiltered and filtered output with a second level ICA procedure. Although the automated filter does not reach performance values of visual analysis by human raters, we propose that resting-state compatible analysis of ICA time courses could be very useful to complement the existing map or task/event oriented artifact classification algorithms.

Automatic selection of resting-state networks with functional magnetic resonance imaging

Functional magnetic resonance imaging (fMRI) during a resting-state condition can reveal the co-activation of specific brain regions in distributed networks, called resting-state networks, which are selected by independent component analysis (ICA) of the fMRI data. One of the major difficulties with component analysis is the automatic selection of the ICA features related to brain activity. In this study we describe a method designed to automatically select networks of potential functional relevance, specifically, those regions known to be involved in motor function, visual processing, executive functioning, auditory processing, memory, and the default-mode network. To do this, image analysis was based on probabilistic ICA as implemented in FSL software. After decomposition, the optimal number of components was selected by applying a novel algorithm which takes into account, for each component, Pearson's median coefficient of skewness of the spatial maps generated by FSL, followed by clustering, segmentation, and spectral analysis. To evaluate the performance of the approach, we investigated the resting-state networks in 25 subjects. For each subject, three resting-state scans were obtained with a Siemens Allegra 3 T scanner (NYU data set). Comparison of the visually and the automatically identified neuronal networks showed that the algorithm had high accuracy (first scan: 95%, second scan: 95%, third scan: 93%) and precision (90%, 90%, 84%). The reproducibility of the networks for visual and automatic selection was very close: it was highly consistent in each subject for the default-mode network (≥92%) and the occipital network, which includes the medial visual cortical areas (≥94%), and consistent for the attention network (≥80%), the right and/or left lateralized frontoparietal attention networks, and the temporal-motor network (≥80%). The automatic selection method may be used to detect neural networks and reduce subjectivity in ICA component assessment.

Neural correlates of behavioral variation in healthy adults' antisaccade performance

Cognitive control is required for correct antisaccade performance. High antisaccade error rates characterize certain psychiatric disorders, but can be highly variable, even among healthy groups. Antisaccade data were acquired from a large sample of healthy undergraduates, and error rate was quantified. Participants who reliably made few errors (good, n = 13) or many errors (poor, n = 13) were recruited back to perform antisaccades during fMRI acquisition. A data-derived model was used to compare signal between good and poor performers during blocks of antisaccade trials. Behaviorally derived regressors were used to compare signal between good and poor performers during correct and error trials. Results show differential activation in middle frontal gyrus and inferior parietal lobule between good and poor performers, suggesting that failure to recruit these top-down control regions corresponds to poor antisaccade performance in healthy young adults.

Spontaneous EEG-Functional MRI in Mesial Temporal Lobe Epilepsy: Implications for the Neural Correlates of Consciousness

The combination of electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) has been shown to have great potential for providing a greater understanding of normal and diseased states in both human and animal studies. Simultaneous EEG-fMRI is particularly well suited for the study of epilepsy in that it may reveal the neurobiology of ictal and interictal epileptiform discharges and noninvasively localize epileptogenic foci. Spontaneous, coherent fluctuations of neuronal activity and the coupled hemodynamic responses have also been shown to provide diagnostic markers of disease, extending our understanding of intrinsically structured ongoing brain activity. Following a short summary of the hardware and software development of simultaneous EEG-fMRI, this paper reviews a unified framework of integrating neuronal and hemodynamic processes during epileptic seizures and discusses the role and impact of spontaneous activity in the mesial temporal lobe epilepsies with particular emphasis on the neural and physiological correlates of consciousness.

Order Selection of the Linear Mixing Model for Complex-valued FMRI Data

Functional magnetic resonance imaging (fMRI) data are originally acquired as complex-valued images, which motivates the use of complex-valued data analysis methods. Due to the high dimension and high noise level of fMRI data, order selection and dimension reduction are important procedures for multivariate analysis methods such as independent component analysis (ICA). In this work, we develop a complex-valued order selection method to estimate the dimension of signal subspace using information-theoretic criteria. To correct the effect of sample dependence to information-theoretic criteria, we develop a general entropy rate measure for complex Gaussian random process to calibrate the independent and identically distributed (i.i.d.) sampling scheme in the complex domain. We show the effectiveness of the approach for order selection on both simulated and actual fMRI data. A comparison between the results of order selection and ICA on real-valued and complex-valued fMRI data demonstrates that a fully complex analysis extracts more information about brain activation.

A data-driven sparse GLM for fMRI analysis using sparse dictionary learning with MDL criterion

We propose a novel statistical analysis method for functional magnetic resonance imaging (fMRI) to overcome the drawbacks of conventional data-driven methods such as the independent component analysis (ICA). Although ICA has been broadly applied to fMRI due to its capacity to separate spatially or temporally independent components, the assumption of independence has been challenged by recent studies showing that ICA does not guarantee independence of simultaneously occurring distinct activity patterns in the brain. Instead, sparsity of the signal has been shown to be more promising. This coincides with biological findings such as sparse coding in V1 simple cells, electrophysiological experiment results in the human medial temporal lobe, etc. The main contribution of this paper is, therefore, a new data driven fMRI analysis that is derived solely based upon the sparsity of the signals. A compressed sensing based data-driven sparse generalized linear model is proposed that enables estimation of spatially adaptive design matrix as well as sparse signal components that represent synchronous, functionally organized and integrated neural hemodynamics. Furthermore, a minimum description length (MDL)-based model order selection rule is shown to be essential in selecting unknown sparsity level for sparse dictionary learning. Using simulation and real fMRI experiments, we show that the proposed method can adapt individual variation better compared to the conventional ICA methods.

Altered functional and anatomical connectivity in schizophrenia

BACKGROUND

Schizophrenia is characterized by a lack of integration between thought, emotion, and behavior. A disruption in the connectivity between brain processes may underlie this schism. Functional magnetic resonance imaging (fMRI) and diffusion tensor imaging (DTI) were used to evaluate functional and anatomical brain connectivity in schizophrenia.

METHODS

In all, 29 chronic schizophrenia patients (11 females, age: mean=41.3, SD=9.28) and 29 controls (11 females, age: mean=41.1, SD=10.6) were recruited. Schizophrenia patients were assessed for severity of negative and positive symptoms and general cognitive abilities of attention/concentration and memory. Participants underwent a resting-fMRI scan and a DTI scan. For fMRI data, a hybrid independent components analysis was used to extract the group default mode network (DMN) and accompanying time-courses. Voxel-wise whole-brain multiple regressions with corresponding DMN time-courses was conducted for each subject. A t-test was conducted on resulting DMN correlation maps to look between-group differences. For DTI data, voxel-wise statistical analysis of the fractional anisotropy data was carried out to look for between-group differences. Voxel-wise correlations were conducted to investigate the relationship between brain connectivity and behavioral measures.

RESULTS

Results revealed altered functional and anatomical connectivity in medial frontal and anterior cingulate gyri of schizophrenia patients. In addition, frontal connectivity in schizophrenia patients was positively associated with symptoms as well as with general cognitive ability measures.

DISCUSSION

The present study shows convergent fMRI and DTI findings that are consistent with the disconnection hypothesis in schizophrenia, particularly in medial frontal regions, while adding some insight of the relationship between brain disconnectivity and behavior.

Automatic Bayesian classification of healthy controls, bipolar disorder, and schizophrenia using intrinsic connectivity maps from FMRI data

We present a method for supervised, automatic, and reliable classification of healthy controls, patients with bipolar disorder, and patients with schizophrenia using brain imaging data. The method uses four supervised classification learning machines trained with a stochastic gradient learning rule based on the minimization of Kullback-Leibler divergence and an optimal model complexity search through posterior probability estimation. Prior to classification, given the high dimensionality of functional MRI (fMRI) data, a dimension reduction stage comprising two steps is performed: first, a one-sample univariate t-test mean-difference T(score) approach is used to reduce the number of significant discriminative functional activated voxels, and then singular value decomposition is performed to further reduce the dimension of the input patterns to a number comparable to the limited number of subjects available for each of the three classes. Experimental results using functional brain imaging (fMRI) data include receiver operation characteristic curves for the three-way classifier with area under curve values around 0.82, 0.89, and 0.90 for healthy control versus nonhealthy, bipolar disorder versus nonbipolar, and schizophrenia patients versus nonschizophrenia binary problems, respectively. The average three-way correct classification rate (CCR) is in the range of 70%-72%, for the test set, remaining close to the estimated Bayesian optimal CCR theoretical upper bound of about 80% , estimated from the one nearest-neighbor classifier over the same data.

Hybrid ICA-Seed-Based Methods for fMRI Functional Connectivity Assessment: A Feasibility Study

Brain functional connectivity (FC) is often assessed from fMRI data using seed-based methods, such as those of detecting temporal correlation between a predefined region (seed) and all other regions in the brain; or using multivariate methods, such as independent component analysis (ICA). ICA is a useful data-driven tool, but reproducibility issues complicate group inferences based on FC maps derived with ICA. These reproducibility issues can be circumvented with hybrid methods that use information from ICA-derived spatial maps as seeds to produce seed-based FC maps. We report results from five experiments to demonstrate the potential advantages of hybrid ICA-seed-based FC methods, comparing results from regressing fMRI data against task-related a priori time courses, with “back-reconstruction” from a group ICA, and with five hybrid ICA-seed-based FC methods: ROI-based with (1) single-voxel, (2) few-voxel, and (3) many-voxel seed; and dual-regression-based with (4) single ICA map and (5) multiple ICA map seed.

Sensory-motor brain network connectivity for speech comprehension

The act of listening to speech activates a large network of brain areas. In the present work, a novel data-driven technique (the combination of independent component analysis and Granger causality) was used to extract brain network dynamics from an fMRI study of passive listening to Words, Pseudo-Words, and Reverse-played words. Using this method we show the functional connectivity modulations among classical language regions (Broca's and Wernicke's areas) and inferior parietal, somatosensory, and motor areas and right cerebellum. Word listening elicited a compact pattern of connectivity within a parieto-somato-motor network and between the superior temporal and inferior frontal gyri. Pseudo-Word stimuli induced activities similar to the Word condition, which were characterized by a highly recurrent connectivity pattern, mostly driven by the temporal lobe activity. Also the Reversed-Word condition revealed an important influence of temporal cortices, but no integrated activity of the parieto-somato-motor network. In parallel, the right cerebellum lost its functional connection with motor areas, present in both Word and Pseudo-Word listening. The inability of the participant to produce the Reversed-Word stimuli also evidenced two separate networks: the first was driven by frontal areas and the right cerebellum toward somatosensory cortices; the second was triggered by temporal and parietal sites towards motor areas. Summing up, our results suggest that semantic content modulates the general compactness of network dynamics as well as the balance between frontal and temporal language areas in driving those dynamics. The degree of reproducibility of auditory speech material modulates the connectivity pattern within and toward somatosensory and motor areas.

Visual inspection of independent components: defining a procedure for artifact removal from fMRI data

Artifacts in functional magnetic resonance imaging (fMRI) data, primarily those related to motion and physiological sources, negatively impact the functional signal-to-noise ratio in fMRI studies, even after conventional fMRI preprocessing. Independent component analysis' demonstrated capacity to separate sources of neural signal, structured noise, and random noise into separate components might be utilized in improved procedures to remove artifacts from fMRI data. Such procedures require a method for labeling independent components (ICs) as representing artifacts to be removed or neural signals of interest to be spared. Visual inspection is often considered an accurate method for such labeling as well as a standard to which automated labeling methods are compared. However, detailed descriptions of methods for visual inspection of ICs are lacking in the literature. Here we describe the details of, and the rationale for, an operationalized fMRI data denoising procedure that involves visual inspection of ICs (96% inter-rater agreement). We estimate that dozens of subjects/sessions can be processed within a few hours using the described method of visual inspection. Our hope is that continued scientific discussion of and testing of visual inspection methods will lead to the development of improved, cost-effective fMRI denoising procedures.

On entropy rate for the complex domain

We derive the entropy rate formula for a complex Gaussian random process by using a widely linear model. The resulting expression is general and applicable to both circular and noncircular Gaussian processes, since any second-order stationary process can be modeled as the output of a widely linear system driven by a circular white noise. Furthermore, we demonstrate application of the derived formula to an order selection problem. We extend a scheme for independent and identically distributed (i.i.d.) sampling to the complex domain to improve the estimation performance of information-theoretic criteria when samples are correlated. We show the effectiveness of the approach for order selection for simulated and actual functional magnetic resonance imaging (fMRI) data that are inherently complex valued.

Focusing effects of L-dopa in Parkinson's disease

Previous fMRI motor studies in Parkinson's disease (PD) have suggested that L-dopa may "normalize" areas of hypo- and hyperactivity. However, results from these studies, which were largely based on analyzing BOLD signal amplitude, have been conflicting. Examining only amplitude changes at distinct loci may thus be inadequate in fully capturing the activation changes induced by L-dopa. In this article, we extended prior analyses on the effects of L-dopa by investigating both amplitude and spatial changes of brain activation before and after L-dopa. Ten subjects with PD, both on and off medication, and ten healthy, age-matched controls performed a visuo-motor tracking task in which they sinusoidally squeezed a bulb at 0.25, 0.5, and 0.75 Hz. This task was contrasted with static squeezing to generate fMRI activation maps. To investigate the effects of L-dopa, we examined the amplitude and spatial variance of the BOLD response within anatomically-defined regions of interest (ROIs). L-dopa had significant main effects on the amplitude of BOLD signal in bilateral primary motor cortex and left SMA. In contrast, L-dopa-mediated spatial changes were apparent in bilateral cerebellar hemispheres, M1, SMA, and right prefrontal cortex. Moreover, L-dopa appeared to normalize the spatial distribution of ROI activation in PD to that of the controls. Specifically, L-dopa had a "focusing" effect on activity-an effect more pronounced than the typically-measured fMRI amplitude changes. This observation is consistent with modeling studies, which demonstrated that dopamine increases the signal-to-noise ratio at the neuronal level with a resultant focusing of representations at the macroscopic level.