A resampling approach to estimate the stability of one-dimensional or multidimensional independent components

Basis profile curve identification to understand electrical stimulation effects in human brain networks

Brain networks can be explored by delivering brief pulses of electrical current in one area while measuring voltage responses in other areas. We propose a convergent paradigm to study brain dynamics, focusing on a single brain site to observe the average effect of stimulating each of many other brain sites. Viewed in this manner, visually-apparent motifs in the temporal response shape emerge from adjacent stimulation sites. This work constructs and illustrates a data-driven approach to determine characteristic spatiotemporal structure in these response shapes, summarized by a set of unique "basis profile curves" (BPCs). Each BPC may be mapped back to underlying anatomy in a natural way, quantifying projection strength from each stimulation site using simple metrics. Our technique is demonstrated for an array of implanted brain surface electrodes in a human patient. This framework enables straightforward interpretation of single-pulse brain stimulation data, and can be applied generically to explore the diverse milieu of interactions that comprise the connectome.

Improving motor imagery classification during induced motor perturbations

Objective.Motor imagery is the mental simulation of movements. It is a common paradigm to design brain-computer interfaces (BCIs) that elicits the modulation of brain oscillatory activity similar to real, passive and induced movements. In this study, we used peripheral stimulation to provoke movements of one limb during the performance of motor imagery tasks. Unlike other works, in which induced movements are used to support the BCI operation, our goal was to test and improve the robustness of motor imagery based BCI systems to perturbations caused by artificially generated movements.Approach.We performed a BCI session with ten participants who carried out motor imagery of three limbs. In some of the trials, one of the arms was moved by neuromuscular stimulation. We analysed 2-class motor imagery classifications with and without movement perturbations. We investigated the performance decrease produced by these disturbances and designed different computational strategies to attenuate the observed classification accuracy drop.Main results.When the movement was induced in a limb not coincident with the motor imagery classes, extracting oscillatory sources of the movement imagination tasks resulted in BCI performance being similar to the control (undisturbed) condition; when the movement was induced in a limb also involved in the motor imagery tasks, the performance drop was significantly alleviated by spatially filtering out the neural noise caused by the stimulation. We also show that the loss of BCI accuracy was accompanied by weaker power of the sensorimotor rhythm. Importantly, this residual power could be used to predict whether a BCI user will perform with sufficient accuracy under the movement disturbances.Significance.We provide methods to ameliorate and even eliminate motor related afferent disturbances during the performance of motor imagery tasks. This can help improving the reliability of current motor imagery based BCI systems.

Towards deep phenotyping pregnancy: a systematic review on artificial intelligence and machine learning methods to improve pregnancy outcomes

Objective

Development of novel informatics methods focused on improving pregnancy outcomes remains an active area of research. The purpose of this study is to systematically review the ways that artificial intelligence (AI) and machine learning (ML), including deep learning (DL), methodologies can inform patient care during pregnancy and improve outcomes.

Materials and methods

We searched English articles on EMBASE, PubMed and SCOPUS. Search terms included ML, AI, pregnancy and informatics. We included research articles and book chapters, excluding conference papers, editorials and notes.

Results

We identified 127 distinct studies from our queries that were relevant to our topic and included in the review. We found that supervised learning methods were more popular (n = 69) than unsupervised methods (n = 9). Popular methods included support vector machines (n = 30), artificial neural networks (n = 22), regression analysis (n = 17) and random forests (n = 16). Methods such as DL are beginning to gain traction (n = 13). Common areas within the pregnancy domain where AI and ML methods were used the most include prenatal care (e.g. fetal anomalies, placental functioning) (n = 73); perinatal care, birth and delivery (n = 20); and preterm birth (n = 13). Efforts to translate AI into clinical care include clinical decision support systems (n = 24) and mobile health applications (n = 9).

Conclusions

Overall, we found that ML and AI methods are being employed to optimize pregnancy outcomes, including modern DL methods (n = 13). Future research should focus on less-studied pregnancy domain areas, including postnatal and postpartum care (n = 2). Also, more work on clinical adoption of AI methods and the ethical implications of such adoption is needed.

Consistency of independent component analysis for FMRI

BACKGROUND

Independent component analysis (ICA) has been widely used for blind source separation in the field of medical imaging. However, despite of previous substantial efforts, the stability of ICA components remains a critical issue which has not been adequately addressed, despite numerous previous efforts. Most critical is the inconsistency of some of the extracted components when ICA is run with different model orders (MOs).

NEW METHOD

In this study, a novel method of determining the consistency of component analysis (CoCA) is proposed to evaluate the consistency of extracted components with different model orders. In the method, "consistent components" (CCs) are defined as those which can be extracted repeatably over a range of model orders.

RESULT

The efficacy of the method was evaluated with simulation data and fMRI datasets. With our method, the simulation result showed a clear difference of consistency between ground truths and noise.

COMPARISON WITH EXISTING METHODS

The information criteria were implemented to provide suggestions for the optimal model order, where some of the ICs were revealed inconsistent in our proposed method.

CONCLUSIONS

This method provided an objective protocol for choosing CCs of an ICA decomposition of a data matrix, independent of model order. This is especially useful with high model orders, where noise or other disturbances could possibly lead to an instability of the components.

Variability of ICA decomposition may impact EEG signals when used to remove eyeblink artifacts

Despite the growing use of independent component analysis (ICA) algorithms for isolating and removing eyeblink-related activity from EEG data, we have limited understanding of how variability associated with ICA uncertainty may be influencing the reconstructed EEG signal after removing the eyeblink artifact components. To characterize the magnitude of this ICA uncertainty and to understand the extent to which it may influence findings within ERP and EEG investigations, ICA decompositions of EEG data from 32 college-aged young adults were repeated 30 times for three popular ICA algorithms. Following each decomposition, eyeblink components were identified and removed. The remaining components were back-projected, and the resulting clean EEG data were further used to analyze ERPs. Findings revealed that ICA uncertainty results in variation in P3 amplitude as well as variation across all EEG sampling points, but differs across ICA algorithms as a function of the spatial location of the EEG channel. This investigation highlights the potential of ICA uncertainty to introduce additional sources of variance when the data are back-projected without artifact components. Careful selection of ICA algorithms and parameters can reduce the extent to which ICA uncertainty may introduce an additional source of variance within ERP/EEG studies.

Cortical network architecture for context processing in primate brain

Context is information linked to a situation that can guide behavior. In the brain, context is encoded by sensory processing and can later be retrieved from memory. How context is communicated within the cortical network in sensory and mnemonic forms is unknown due to the lack of methods for high-resolution, brain-wide neuronal recording and analysis. Here, we report the comprehensive architecture of a cortical network for context processing. Using hemisphere-wide, high-density electrocorticography, we measured large-scale neuronal activity from monkeys observing videos of agents interacting in situations with different contexts. We extracted five context-related network structures including a bottom-up network during encoding and, seconds later, cue-dependent retrieval of the same network with the opposite top-down connectivity. These findings show that context is represented in the cortical network as distributed communication structures with dynamic information flows. This study provides a general methodology for recording and analyzing cortical network neuronal communication during cognition.

DOI:http://dx.doi.org/10.7554/eLife.06121.001

If we see someone looking frightened, the way we respond to the situation is influenced by other information, referred to as the ‘context’. For example, if the person is frightened because another individual is shouting at them, we might try to intervene. However, if the person is watching a horror video we may decide that they don't need our help and leave them to it. Nevertheless, it is not clear how the brain processes the context of a situation to inform our response.

Here, Chao et al. developed a new method to study electrical activity across the whole of the brain and used it to study how monkeys process context in response to several different social situations. In the experiments, monkeys were shown video clips in which one monkey—known as the ‘video monkey’—was threatened by a human or another monkey, or in which the video monkey is facing an empty wall (i.e., in three different contexts). Afterwards, the video monkey either displays a frightened expression or a neutral one. Chao et al. found that if the video monkey looked frightened by the context, the monkeys watching the video clip shifted their gaze to observe the apparent threat. How these monkeys shifted their gaze depended on the context, but this behavior was absent when the video monkey gave a neutral expression.

The experiments used an array of electrodes that covered a wide area of the monkeys' brains to record electrical activity of nerve cells as the monkeys watched the videos. Chao et al. investigated how brain regions communicated with each other in response to different contexts, and found that the information of contexts was presented in the interactions between distant brain regions. The monkeys' brains sent information from a region called the temporal cortex (which is involved in processing sensory and social information), to another region called the prefrontal cortex (which is involved in functions such as reasoning, attention, and memory). Seconds later, the flow of information was reversed as the monkeys utilized information about the context to guide their behavior.

Chao et al.'s findings reveal how information about the context of a situation is transmitted around the brain to inform a response. The next challenge is to experimentally manipulate the identified brain circuits to investigate if problems in context processing could lead to the inappropriate responses that contribute to schizophrenia, post-traumatic stress disorder and other psychiatric disorders.

DOI:http://dx.doi.org/10.7554/eLife.06121.002

RELICA: a method for estimating the reliability of independent components

Independent Component Analysis (ICA) is a widely applied data-driven method for parsing brain and non-brain EEG source signals, mixed by volume conduction to the scalp electrodes, into a set of maximally temporally and often functionally independent components (ICs). Many ICs may be identified with a precise physiological or non-physiological origin. However, this process is hindered by partial instability in ICA results that can arise from noise in the data. Here we propose RELICA (RELiable ICA), a novel method to characterize IC reliability within subjects. RELICA first computes IC "dipolarity" a measure of physiological plausibility, plus a measure of IC consistency across multiple decompositions of bootstrap versions of the input data. RELICA then uses these two measures to visualize and cluster the separated ICs, providing a within-subject measure of IC reliability that does not involve checking for its occurrence across subjects. We demonstrate the use of RELICA on EEG data recorded from 14 subjects performing a working memory experiment and show that many brain and ocular artifact ICs are correctly classified as "stable" (highly repeatable across decompositions of bootstrapped versions of the input data). Many stable ICs appear to originate in the brain, while other stable ICs account for identifiable non-brain processes such as line noise. RELICA might be used with any linear blind source separation algorithm to reduce the risk of basing conclusions on unstable or physiologically un-interpretable component processes.

Second order blind identification on the cerebral cortex

Blind source separation (BSS) methods have become standard brain imaging tools and are routinely used for noise and artifact removal, as well as for extracting related spatial and temporal components from brain signals. Despite their popularity, BSS methods have rarely been used to explore maps of cortical thickness and sulcal folding patterns. Our limited knowledge of the relationship between cortical morphometry, brain development and pathologies of the central nervous system makes BSS methods ideal investigative tools. We propose a novel spatial BSS method tailored for application to the cerebral cortex based on the second order blind identification (SOBI) method. Our method outperforms the regular SOBI and popular FastICA BSS methods on simulations. Application to maps of cortical thickness and curvature from normal controls reveals original structural networks.

Independent component analysis: recent advances

Independent component analysis is a probabilistic method for learning a linear transform of a random vector. The goal is to find components that are maximally independent and non-Gaussian (non-normal). Its fundamental difference to classical multi-variate statistical methods is in the assumption of non-Gaussianity, which enables the identification of original, underlying components, in contrast to classical methods. The basic theory of independent component analysis was mainly developed in the 1990s and summarized, for example, in our monograph in 2001. Here, we provide an overview of some recent developments in the theory since the year 2000. The main topics are: analysis of causal relations, testing independent components, analysing multiple datasets (three-way data), modelling dependencies between the components and improved methods for estimating the basic model.

Repeated decompositions reveal the stability of infomax decomposition of fMRI data

In this study, we decomposed 12 fMRI data sets from six subjects each 101 times using the infomax algorithm. The first decomposition was taken as a reference decomposition; the others were used to form a component matrix of 100 by 100 components. Equivalence relations between components in this matrix, defined as maximum spatial correlations to the components of the reference decomposition, were found by the Hungarian sorting method and used to form 100 equivalence classes for each data set. We then tested the reproducibility of the matched components in the equivalence classes using uncertainty measures based on component distributions, time courses, and ROC curves. Infomax ICA rarely failed to derive nearly the same components in different decompositions. Very few components per data set were poorly reproduced, even using vector angle uncertainty measures stricter than correlation and detection theory measures.

Identifying reliable independent components via split-half comparisons

Independent component analysis (ICA) is a family of unsupervised learning algorithms that have proven useful for the analysis of the electroencephalogram (EEG) and magnetoencephalogram (MEG). ICA decomposes an EEG/MEG data set into a basis of maximally temporally independent components (ICs) that are learned from the data. As with any statistic, a concern with using ICA is the degree to which the estimated ICs are reliable. An IC may not be reliable if ICA was trained on insufficient data, if ICA training was stopped prematurely or at a local minimum (for some algorithms), or if multiple global minima were present. Consequently, evidence of ICA reliability is critical for the credibility of ICA results. In this paper, we present a new algorithm for assessing the reliability of ICs based on applying ICA separately to split-halves of a data set. This algorithm improves upon existing methods in that it considers both IC scalp topographies and activations, uses a probabilistically interpretable threshold for accepting ICs as reliable, and requires applying ICA only three times per data set. As evidence of the method's validity, we show that the method can perform comparably to more time intensive bootstrap resampling and depends in a reasonable manner on the amount of training data. Finally, using the method we illustrate the importance of checking the reliability of ICs by demonstrating that IC reliability is dramatically increased by removing the mean EEG at each channel for each epoch of data rather than the mean EEG in a prestimulus baseline.

Dissociating functional brain networks by decoding the between-subject variability

In this study we illustrate how the functional networks involved in a single task (e.g. the sensory, cognitive and motor components) can be segregated without cognitive subtractions at the second-level. The method used is based on meaningful variability in the patterns of activation between subjects with the assumption that regions belonging to the same network will have comparable variations from subject to subject. fMRI data were collected from thirty nine healthy volunteers who were asked to indicate with a button press if visually presented words were semantically related or not. Voxels were classified according to the similarity in their patterns of between-subject variance using a second-level unsupervised fuzzy clustering algorithm. The results were compared to those identified by cognitive subtractions of multiple conditions tested in the same set of subjects. This illustrated that the second-level clustering approach (on activation for a single task) was able to identify the functional networks observed using cognitive subtractions (e.g. those associated with vision, semantic associations or motor processing). In addition the fuzzy clustering approach revealed other networks that were not dissociated by the cognitive subtraction approach (e.g. those associated with high- and low-level visual processing and oculomotor movements). We discuss the potential applications of our method which include the identification of “hidden” or unpredicted networks as well as the identification of systems level signatures for different subgroupings of clinical and healthy populations.

A hybrid technique for blind separation of non-gaussian and time-correlated sources using a multicomponent approach

Blind inversion of a linear and instantaneous mixture of source signals is a problem often encountered in many signal processing applications. Efficient fastICA (EFICA) offers an asymptotically optimal solution to this problem when all of the sources obey a generalized Gaussian distribution, at most one of them is Gaussian, and each is independent and identically distributed (i.i.d.) in time. Likewise, weights-adjusted second-order blind identification (WASOBI) is asymptotically optimal when all the sources are Gaussian and can be modeled as autoregressive (AR) processes with distinct spectra. Nevertheless, real-life mixtures are likely to contain both Gaussian AR and non-Gaussian i.i.d. sources, rendering WASOBI and EFICA severely suboptimal. In this paper, we propose a novel scheme for combining the strengths of EFICA and WASOBI in order to deal with such hybrid mixtures. Simulations show that our approach outperforms competing algorithms designed for separating similar mixtures.

Estimating the number of independent components for functional magnetic resonance imaging data

Multivariate analysis methods such as independent component analysis (ICA) have been applied to the analysis of functional magnetic resonance imaging (fMRI) data to study brain function. Because of the high dimensionality and high noise level of the fMRI data, order selection, i.e., estimation of the number of informative components, is critical to reduce over/underfitting in such methods. Dependence among fMRI data samples in the spatial and temporal domain limits the usefulness of the practical formulations of information-theoretic criteria (ITC) for order selection, since they are based on likelihood of independent and identically distributed (i.i.d.) data samples. To address this issue, we propose a subsampling scheme to obtain a set of effectively i.i.d. samples from the dependent data samples and apply the ITC formulas to the effectively i.i.d. sample set for order selection. We apply the proposed method on the simulated data and show that it significantly improves the accuracy of order selection from dependent data. We also perform order selection on fMRI data from a visuomotor task and show that the proposed method alleviates the over-estimation on the number of brain sources due to the intrinsic smoothness and the smooth preprocessing of fMRI data. We use the software package ICASSO (Himberg et al. [ 2004]: Neuroimage 22:1214-1222) to analyze the independent component (IC) estimates at different orders and show that, when ICA is performed at overestimated orders, the stability of the IC estimates decreases and the estimation of task related brain activations show degradation.

The selection of optimal ICA algorithm parameters for robust AEP component estimates using 3 popular ICA algorithms

Many authors have used the Auditory Evoked Potential (AEP) recordings to evaluate the performance of their ICA algorithms and have demonstrated that this procedure can remove the typical EEG artifact in these recordings (i.e. blinking, muscle noise, line noise, etc.). However, there is little work in the literature about the optimal parameters, for each of those algorithms, for the estimation of the AEP components to reliably recover both the auditory response and the specific artifacts generated for the normal function of a Cochlear Implant (CI), used for the rehabilitation of deaf people. In this work we determine the optimal parameters of three ICA algorithms, each based on different independence criteria, and assess the resulting estimations of both the auditory response and CI artifact. We show that the algorithm utilizing temporal structure, such as TDSEP-ICA, is better in estimating the components of the auditory response, in recordings contaminated by CI artifacts, than higher order statistics based algorithms.

Analyzing consistency of independent components: an fMRI illustration

Independent component analysis (ICA) is a powerful data-driven signal processing technique. It has proved to be helpful in, e.g., biomedicine, telecommunication, finance and machine vision. Yet, some problems persist in its wider use. One concern is the reliability of solutions found with ICA algorithms, resulting from the stochastic changes each time the analysis is performed. The consistency of the solutions can be analyzed by clustering solutions from multiple runs of bootstrapped ICA. Related methods have been recently published either for analyzing algorithmic stability or reducing the variability. The presented approach targets the extraction of additional information related to the independent components, by focusing on the nature of the variability. Practical implications are illustrated through a functional magnetic resonance imaging (fMRI) experiment.

Fetal auditory responses to external sounds and mother's heart beat: detection improved by Independent Component Analysis

In this paper, we present a magnetoencephalographic study of the fetal auditory response to external stimuli and to the sound of the mother's heartbeat. We describe how an ad hoc functional selection procedure allowed us to isolate the sources in the fetal brain responding to sounds only, after the application to the recorded data of a standard Independent Component Analysis algorithm. In our experiment, acoustic stimuli were delivered to twelve healthy women with uncomplicated pregnancies at a time between 36 and 40 weeks gestational age, with their fetuses in breech presentation. Ultrasound images allowed determination of the region over the women's abdomen nearest to the fetal head, over which both the acoustic stimulator and the MEG sensors were subsequently placed. In 8 out of the 12 cases, our analysis provided consistent evidence of a fetal response both to the mother's heartbeat and to the external auditory stimulation; both were characterized by a clear prominent component at around 200 ms latency, which is widely accepted as the marker of the fetal response to auditory stimuli.

Enhancing the signal-to-noise ratio of ICA-based extracted ERPs

When decomposing single trial electroencephalography it is a challenge to incorporate prior physiological knowledge. Here, we develop a method that uses prior information about the phase-locking property of event-related potentials in a regularization framework to bias a blind source separation algorithm toward an improved separation of single-trial phase-locked responses in terms of an increased signal-to-noise ratio. In particular, we suggest a transformation of the data, using weighted average of the single trial and trial-averaged response, that redirects the focus of source separation methods onto the subspace of event-related potentials. The practical benefit with respect to an improved separation of such components from ongoing background activity and extraneous noise is first illustrated on artificial data and finally verified in a real-world application of extracting single-trial somatosensory evoked potentials from multichannel EEG-recordings.

Least-dependent-component analysis based on mutual information

We propose to use precise estimators of mutual information (MI) to find the least dependent components in a linearly mixed signal. On the one hand, this seems to lead to better blind source separation than with any other presently available algorithm. On the other hand, it has the advantage, compared to other implementations of "independent" component analysis (ICA), some of which are based on crude approximations for MI, that the numerical values of the MI can be used for (i) estimating residual dependencies between the output components; (ii) estimating the reliability of the output by comparing the pairwise MIs with those of remixed components; and (iii) clustering the output according to the residual interdependencies. For the MI estimator, we use a recently proposed k -nearest-neighbor-based algorithm. For time sequences, we combine this with delay embedding, in order to take into account nontrivial time correlations. After several tests with artificial data, we apply the resulting MILCA (mutual-information-based least dependent component analysis) algorithm to a real-world dataset, the ECG of a pregnant woman.

Validating the independent components of neuroimaging time series via clustering and visualization

Recently, independent component analysis (ICA) has been widely used in the analysis of brain imaging data. An important problem with most ICA algorithms is, however, that they are stochastic; that is, their results may be somewhat different in different runs of the algorithm. Thus, the outputs of a single run of an ICA algorithm should be interpreted with some reserve, and further analysis of the algorithmic reliability of the components is needed. Moreover, as with any statistical method, the results are affected by the random sampling of the data, and some analysis of the statistical significance or reliability should be done as well. Here we present a method for assessing both the algorithmic and statistical reliability of estimated independent components. The method is based on running the ICA algorithm many times with slightly different conditions and visualizing the clustering structure of the obtained components in the signal space. In experiments with magnetoencephalographic (MEG) and functional magnetic resonance imaging (fMRI) data, the method was able to show that expected components are reliable; furthermore, it pointed out components whose interpretation was not obvious but whose reliability should incite the experimenter to investigate the underlying technical or physical phenomena. The method is implemented in a software package called Icasso.

Estimating mutual information

We present two classes of improved estimators for mutual information M(X,Y), from samples of random points distributed according to some joint probability density mu(x,y). In contrast to conventional estimators based on binnings, they are based on entropy estimates from k -nearest neighbor distances. This means that they are data efficient (with k=1 we resolve structures down to the smallest possible scales), adaptive (the resolution is higher where data are more numerous), and have minimal bias. Indeed, the bias of the underlying entropy estimates is mainly due to nonuniformity of the density at the smallest resolved scale, giving typically systematic errors which scale as functions of k/N for N points. Numerically, we find that both families become exact for independent distributions, i.e. the estimator M(X,Y) vanishes (up to statistical fluctuations) if mu(x,y)=mu(x)mu(y). This holds for all tested marginal distributions and for all dimensions of x and y. In addition, we give estimators for redundancies between more than two random variables. We compare our algorithms in detail with existing algorithms. Finally, we demonstrate the usefulness of our estimators for assessing the actual independence of components obtained from independent component analysis (ICA), for improving ICA, and for estimating the reliability of blind source separation.

Bayesian Class Discovery in Microarray Datasets

A novel approach to class discovery in gene expression datasets is presented. In the context of clinical diagnosis, the central goal of class discovery algorithms is to simultaneously find putative (sub-)types of diseases and to identify informative subsets of genes with disease-type specific expression profile. Contrary to many other approaches in the literature, the method presented implements a wrapper strategy for feature selection, in the sense that the features are directly selected by optimizing the discriminative power of the used partitioning algorithm. The usual combinatorial problems associated with wrapper approaches are overcome by a Bayesian inference mechanism. On the technical side, we present an efficient optimization algorithm with guaranteed local convergence property. The only free parameter of the optimization method is selected by a resampling-based stability analysis. Experiments with Leukemia and Lymphoma datasets demonstrate that our method is able to correctly infer partitions and corresponding subsets of genes which both are relevant in a biological sense. Moreover, the frequently observed problem of ambiguities caused by different but equally high-scoring partitions is successfully overcome by the model selection method proposed.