Machine Learning for Automatic Prediction of the Quality of Electrophysiological Recordings

The quality of electrophysiological recordings varies a lot due to technical and biological variability and neuroscientists inevitably have to select “good” recordings for further analyses. This procedure is time-consuming and prone to selection biases. Here, we investigate replacing human decisions by a machine learning approach. We define 16 features, such as spike height and width, select the most informative ones using a wrapper method and train a classifier to reproduce the judgement of one of our expert electrophysiologists. Generalisation performance is then assessed on unseen data, classified by the same or by another expert. We observe that the learning machine can be equally, if not more, consistent in its judgements as individual experts amongst each other. Best performance is achieved for a limited number of informative features; the optimal feature set being different from one data set to another. With 80–90% of correct judgements, the performance of the system is very promising within the data sets of each expert but judgments are less reliable when it is used across sets of recordings from different experts. We conclude that the proposed approach is relevant to the selection of electrophysiological recordings, provided parameters are adjusted to different types of experiments and to individual experimenters.     

Integrating fMRI and SNP data for biomarker identification for schizophrenia with a sparse representation based variable selection method

Background

In recent years, both single-nucleotide polymorphism (SNP) array and functional magnetic resonance imaging (fMRI) have been widely used for the study of schizophrenia (SCZ). In addition, a few studies have been reported integrating both SNPs data and fMRI data for comprehensive analysis.

Methods

In this study, a novel sparse representation based variable selection (SRVS) method has been proposed and tested on a simulation data set to demonstrate its multi-resolution properties. Then the SRVS method was applied to an integrative analysis of two different SCZ data sets, a Single-nucleotide polymorphism (SNP) data set and a functional resonance imaging (fMRI) data set, including 92 cases and 116 controls. Biomarkers for the disease were identified and validated with a multivariate classification approach followed by a leave one out (LOO) cross-validation. Then we compared the results with that of a previously reported sparse representation based feature selection method.

Results

Results showed that biomarkers from our proposed SRVS method gave significantly higher classification accuracy in discriminating SCZ patients from healthy controls than that of the previous reported sparse representation method. Furthermore, using biomarkers from both data sets led to better classification accuracy than using single type of biomarkers, which suggests the advantage of integrative analysis of different types of data.

Conclusions

The proposed SRVS algorithm is effective in identifying significant biomarkers for complicated disease as SCZ. Integrating different types of data (e.g. SNP and fMRI data) may identify complementary biomarkers benefitting the diagnosis accuracy of the disease.

     

Predicting future clinical changes of MCI patients using longitudinal and multimodal biomarkers

Accurate prediction of clinical changes of mild cognitive impairment (MCI) patients, including both qualitative change (i.e., conversion to Alzheimer's disease (AD)) and quantitative change (i.e., cognitive scores) at future time points, is important for early diagnosis of AD and for monitoring the disease progression. In this paper, we propose to predict future clinical changes of MCI patients by using both baseline and longitudinal multimodality data. To do this, we first develop a longitudinal feature selection method to jointly select brain regions across multiple time points for each modality. Specifically, for each time point, we train a sparse linear regression model by using the imaging data and the corresponding clinical scores, with an extra 'group regularization' to group the weights corresponding to the same brain region across multiple time points together and to allow for selection of brain regions based on the strength of multiple time points jointly. Then, to further reflect the longitudinal changes on the selected brain regions, we extract a set of longitudinal features from the original baseline and longitudinal data. Finally, we combine all features on the selected brain regions, from different modalities, for prediction by using our previously proposed multi-kernel SVM. We validate our method on 88 ADNI MCI subjects, with both MRI and FDG-PET data and the corresponding clinical scores (i.e., MMSE and ADAS-Cog) at 5 different time points. We first predict the clinical scores (MMSE and ADAS-Cog) at 24-month by using the multimodality data at previous time points, and then predict the conversion of MCI to AD by using the multimodality data at time points which are at least 6-month ahead of the conversion. The results on both sets of experiments show that our proposed method can achieve better performance in predicting future clinical changes of MCI patients than the conventional methods.     

Analysis of activity in fMRI data using affinity propagation clustering

Clustering analysis is a promising data-driven method for the analysis of functional magnetic resonance imaging (fMRI) data. The huge computation load, however, makes it difficult for the practical use. We use affinity propagation clustering (APC), a new clustering algorithm especially for large data sets to detect brain functional activation from fMRI. It considers all data points as possible exemplars through the minimisation of an energy function and message-passing architecture, and obtains the optimal set of exemplars and their corresponding clusters. Four simulation studies and three in vivo fMRI studies reveal that brain functional activation can be effectively detected and that different response patterns can be distinguished using this method. Our results demonstrate that APC is superior to the k-centres clustering, as revealed by their performance measures in the weighted Jaccard coefficient and average squared error. These results suggest that the proposed APC will be useful in detecting brain functional activation from fMRI data.     

Multidimensional analysis and detection of informative features in human brain white matter

The white matter contains long-range connections between different brain regions and the organization of these connections holds important implications for brain function in health and disease. Tractometry uses diffusion-weighted magnetic resonance imaging (dMRI) to quantify tissue properties along the trajectories of these connections. Statistical inference from tractometry usually either averages these quantities along the length of each fiber bundle or computes regression models separately for each point along every one of the bundles. These approaches are limited in their sensitivity, in the former case, or in their statistical power, in the latter. We developed a method based on the sparse group lasso (SGL) that takes into account tissue properties along all of the bundles and selects informative features by enforcing both global and bundle-level sparsity. We demonstrate the performance of the method in two settings: i) in a classification setting, patients with amyotrophic lateral sclerosis (ALS) are accurately distinguished from matched controls. Furthermore, SGL identifies the corticospinal tract as important for this classification, correctly finding the parts of the white matter known to be affected by the disease. ii) In a regression setting, SGL accurately predicts “brain age.” In this case, the weights are distributed throughout the white matter indicating that many different regions of the white matter change over the lifespan. Thus, SGL leverages the multivariate relationships between diffusion properties in multiple bundles to make accurate phenotypic predictions while simultaneously discovering the most relevant features of the white matter.     

Extracting Multiscale Pattern Information of fMRI Based Functional Brain Connectivity with Application on Classification of Autism Spectrum Disorders

We employed a multi-scale clustering methodology known as “data cloud geometry” to extract functional connectivity patterns derived from functional magnetic resonance imaging (fMRI) protocol. The method was applied to correlation matrices of 106 regions of interest (ROIs) in 29 individuals with autism spectrum disorders (ASD), and 29 individuals with typical development (TD) while they completed a cognitive control task. Connectivity clustering geometry was examined at both “fine” and “coarse” scales. At the coarse scale, the connectivity clustering geometry produced 10 valid clusters with a coherent relationship to neural anatomy. A supervised learning algorithm employed fine scale information about clustering motif configurations and prevalence, and coarse scale information about intra- and inter-regional connectivity; the algorithm correctly classified ASD and TD participants with sensitivity of and specificity of . Most of the predictive power of the logistic regression model resided at the level of the fine-scale clustering geometry, suggesting that cellular versus systems level disturbances are more prominent in individuals with ASD. This article provides validation for this multi-scale geometric approach to extracting brain functional connectivity pattern information and for its use in classification of ASD.     

Insights into Colon Cancer Etiology via a Regularized Approach to Gene Set Analysis of GWAS Data

Genome-wide association studies (GWAS) have successfully identified susceptibility loci from marginal association analysis of SNPs. Valuable insight into genetic variation underlying complex diseases will likely be gained by considering functionally related sets of genes simultaneously. One approach is to further develop gene set enrichment analysis methods, which are initiated in gene expression studies, to account for the distinctive features of GWAS data. These features include the large number of SNPs per gene, the modest and sparse SNP associations, and the additional information provided by linkage disequilibrium (LD) patterns within genes. We propose a “gene set ridge regression in association studies (GRASS)” algorithm. GRASS summarizes the genetic structure for each gene as eigenSNPs and uses a novel form of regularized regression technique, termed group ridge regression, to select representative eigenSNPs for each gene and assess their joint association with disease risk. Compared with existing methods, the proposed algorithm greatly reduces the high dimensionality of GWAS data while still accounting for multiple hits and/or LD in the same gene. We show by simulation that this algorithm performs well in situations in which there are a large number of predictors compared to sample size. We applied the GRASS algorithm to a genome-wide association study of colon cancer and identified nicotinate and nicotinamide metabolism and transforming growth factor beta signaling as the top two significantly enriched pathways. Elucidating the role of variation in these pathways may enhance our understanding of colon cancer etiology.     

Analysis of activity in fMRI data using affinity propagation clustering

Clustering analysis is a promising data-driven method for the analysis of functional magnetic resonance imaging (fMRI) data. The huge computation load, however, makes it difficult for the practical use. We use affinity propagation clustering (APC), a new clustering algorithm especially for large data sets to detect brain functional activation from fMRI. It considers all data points as possible exemplars through the minimisation of an energy function and message-passing architecture, and obtains the optimal set of exemplars and their corresponding clusters. Four simulation studies and three in vivo fMRI studies reveal that brain functional activation can be effectively detected and that different response patterns can be distinguished using this method. Our results demonstrate that APC is superior to the k-centres clustering, as revealed by their performance measures in the weighted Jaccard coefficient and average squared error. These results suggest that the proposed APC will be useful in detecting brain functional activation from fMRI data.     

Network analysis of intrinsic functional brain connectivity in Alzheimer's disease

Functional brain networks detected in task-free (“resting-state”) functional magnetic resonance imaging (fMRI) have a small-world architecture that reflects a robust functional organization of the brain. Here, we examined whether this functional organization is disrupted in Alzheimer's disease (AD). Task-free fMRI data from 21 AD subjects and 18 age-matched controls were obtained. Wavelet analysis was applied to the fMRI data to compute frequency-dependent correlation matrices. Correlation matrices were thresholded to create 90-node undirected-graphs of functional brain networks. Small-world metrics (characteristic path length and clustering coefficient) were computed using graph analytical methods. In the low frequency interval 0.01 to 0.05 Hz, functional brain networks in controls showed small-world organization of brain activity, characterized by a high clustering coefficient and a low characteristic path length. In contrast, functional brain networks in AD showed loss of small-world properties, characterized by a significantly lower clustering coefficient (p<0.01), indicative of disrupted local connectivity. Clustering coefficients for the left and right hippocampus were significantly lower (p<0.01) in the AD group compared to the control group. Furthermore, the clustering coefficient distinguished AD participants from the controls with a sensitivity of 72% and specificity of 78%. Our study provides new evidence that there is disrupted organization of functional brain networks in AD. Small-world metrics can characterize the functional organization of the brain in AD, and our findings further suggest that these network measures may be useful as an imaging-based biomarker to distinguish AD from healthy aging.     

Ready…Go: Amplitude of the fMRI Signal Encodes Expectation of Cue Arrival Time

What happens when the brain awaits a signal of uncertain arrival time, as when a sprinter waits for the starting pistol? And what happens just after the starting pistol fires? Using functional magnetic resonance imaging (fMRI), we have discovered a novel correlate of temporal expectations in several brain regions, most prominently in the supplementary motor area (SMA). Contrary to expectations, we found little fMRI activity during the waiting period; however, a large signal appears after the “go” signal, the amplitude of which reflects learned expectations about the distribution of possible waiting times. Specifically, the amplitude of the fMRI signal appears to encode a cumulative conditional probability, also known as the cumulative hazard function. The fMRI signal loses its dependence on waiting time in a “countdown” condition in which the arrival time of the go cue is known in advance, suggesting that the signal encodes temporal probabilities rather than simply elapsed time. The dependence of the signal on temporal expectation is present in “no-go” conditions, demonstrating that the effect is not a consequence of motor output. Finally, the encoding is not dependent on modality, operating in the same manner with auditory or visual signals. This finding extends our understanding of the relationship between temporal expectancy and measurable neural signals.     

Ready…Go: Amplitude of the fMRI Signal Encodes Expectation of Cue Arrival Time

What happens when the brain awaits a signal of uncertain arrival time, as when a sprinter waits for the starting pistol? And what happens just after the starting pistol fires? Using functional magnetic resonance imaging (fMRI), we have discovered a novel correlate of temporal expectations in several brain regions, most prominently in the supplementary motor area (SMA). Contrary to expectations, we found little fMRI activity during the waiting period; however, a large signal appears after the “go” signal, the amplitude of which reflects learned expectations about the distribution of possible waiting times. Specifically, the amplitude of the fMRI signal appears to encode a cumulative conditional probability, also known as the cumulative hazard function. The fMRI signal loses its dependence on waiting time in a “countdown” condition in which the arrival time of the go cue is known in advance, suggesting that the signal encodes temporal probabilities rather than simply elapsed time. The dependence of the signal on temporal expectation is present in “no-go” conditions, demonstrating that the effect is not a consequence of motor output. Finally, the encoding is not dependent on modality, operating in the same manner with auditory or visual signals. This finding extends our understanding of the relationship between temporal expectancy and measurable neural signals.     

Modeling the Spatial and Temporal Dependence in fMRI Data

Summary Functional magnetic resonance imaging (fMRI) data sets are large and characterized by complex dependence structures driven by highly sophisticated neurophysiology and aspects of the experimental designs. Typical analyses investigating task‐related changes in measured brain activity use a two‐stage procedure in which the first stage involves subject‐specific models and the second‐stage specifies group (or population) level parameters. Customarily, the first‐level accounts for temporal correlations between the serial scans acquired during one scanning session. Despite accounting for these correlations, fMRI studies often include multiple sessions and temporal dependencies may persist between the corresponding estimates of mean neural activity. Further, spatial correlations between brain activity measurements in different locations are often unaccounted for in statistical modeling and estimation. We propose a two‐stage, spatio‐temporal, autoregressive model that simultaneously accounts for spatial dependencies between voxels within the same anatomical region and for temporal dependencies between a subject's estimates from multiple sessions. We develop an algorithm that leverages the special structure of our covariance model, enabling relatively fast and efficient estimation. Using our proposed method, we analyze fMRI data from a study of inhibitory control in cocaine addicts.     

Smoothness without Smoothing: Why Gaussian Naive Bayes Is Not Naive for Multi-Subject Searchlight Studies

Spatial smoothness is helpful when averaging fMRI signals across multiple subjects, as it allows different subjects'' corresponding brain areas to be pooled together even if they are slightly misaligned. However, smoothing is usually not applied when performing multivoxel pattern-based analyses (MVPA), as it runs the risk of blurring away the information that fine-grained spatial patterns contain. It would therefore be desirable, if possible, to carry out pattern-based analyses which take unsmoothed data as their input but which produce smooth images as output. We show here that the Gaussian Naive Bayes (GNB) classifier does precisely this, when it is used in “searchlight” pattern-based analyses. We explain why this occurs, and illustrate the effect in real fMRI data. Moreover, we show that analyses using GNBs produce results at the multi-subject level which are statistically robust, neurally plausible, and which replicate across two independent data sets. By contrast, SVM classifiers applied to the same data do not generate a replication, even if the SVM-derived searchlight maps have smoothing applied to them. An additional advantage of GNB classifiers for searchlight analyses is that they are orders of magnitude faster to compute than more complex alternatives such as SVMs. Collectively, these results suggest that Gaussian Naive Bayes classifiers may be a highly non-naive choice for multi-subject pattern-based fMRI studies.     

Functional and Structural Correlates of Motor Speed in the Cerebellar Anterior Lobe

In athletics, motor performance is determined by different abilities such as technique, endurance, strength and speed. Based on animal studies, motor speed is thought to be encoded in the basal ganglia, sensorimotor cortex and the cerebellum. The question arises whether there is a unique structural feature in the human brain, which allows “power athletes” to perform a simple foot movement significantly faster than “endurance athletes”. We acquired structural and functional brain imaging data from 32 track-and-field athletes. The study comprised of 16 “power athletes” requiring high speed foot movements (sprinters, jumpers, throwers) and 16 endurance athletes (distance runners) which in contrast do not require as high speed foot movements. Functional magnetic resonance imaging (fMRI) was used to identify speed specific regions of interest in the brain during fast and slow foot movements. Anatomical MRI scans were performed to assess structural grey matter volume differences between athletes groups (voxel based morphometry). We tested maximum movement velocity of plantarflexion (PF-V_max) and acquired electromyographical activity of the lateral and medial gastrocnemius muscle. Behaviourally, a significant difference between the two groups of athletes was noted in PF-V_max and fMRI indicates that fast plantarflexions are accompanied by increased activity in the cerebellar anterior lobe. The same region indicates increased grey matter volume for the power athletes compared to the endurance counterparts. Our results suggest that speed-specific neuro-functional and -structural differences exist between power and endurance athletes in the peripheral and central nervous system.     

Eigenvector Centrality Mapping for Analyzing Connectivity Patterns in fMRI Data of the Human Brain

Functional magnetic resonance data acquired in a task-absent condition (“resting state”) require new data analysis techniques that do not depend on an activation model. In this work, we introduce an alternative assumption- and parameter-free method based on a particular form of node centrality called eigenvector centrality. Eigenvector centrality attributes a value to each voxel in the brain such that a voxel receives a large value if it is strongly correlated with many other nodes that are themselves central within the network. Google''s PageRank algorithm is a variant of eigenvector centrality. Thus far, other centrality measures - in particular “betweenness centrality” - have been applied to fMRI data using a pre-selected set of nodes consisting of several hundred elements. Eigenvector centrality is computationally much more efficient than betweenness centrality and does not require thresholding of similarity values so that it can be applied to thousands of voxels in a region of interest covering the entire cerebrum which would have been infeasible using betweenness centrality. Eigenvector centrality can be used on a variety of different similarity metrics. Here, we present applications based on linear correlations and on spectral coherences between fMRI times series. This latter approach allows us to draw conclusions of connectivity patterns in different spectral bands. We apply this method to fMRI data in task-absent conditions where subjects were in states of hunger or satiety. We show that eigenvector centrality is modulated by the state that the subjects were in. Our analyses demonstrate that eigenvector centrality is a computationally efficient tool for capturing intrinsic neural architecture on a voxel-wise level.     

Performance of Feature Selection Methods

High-throughput biological technologies offer the promise of finding feature sets to serve as biomarkers for medical applications; however, the sheer number of potential features (genes, proteins, etc.) means that there needs to be massive feature selection, far greater than that envisioned in the classical literature. This paper considers performance analysis for feature-selection algorithms from two fundamental perspectives: How does the classification accuracy achieved with a selected feature set compare to the accuracy when the best feature set is used and what is the optimal number of features that should be used? The criteria manifest themselves in several issues that need to be considered when examining the efficacy of a feature-selection algorithm: (1) the correlation between the classifier errors for the selected feature set and the theoretically best feature set; (2) the regressions of the aforementioned errors upon one another; (3) the peaking phenomenon, that is, the effect of sample size on feature selection; and (4) the analysis of feature selection in the framework of high-dimensional models corresponding to high-throughput data.     

FMRI Effective Connectivity and TMS Chronometry: Complementary Accounts of Causality in the Visuospatial Judgment Network

Background

While traditionally quite distinct, functional neuroimaging (e.g. functional magnetic resonance imaging: fMRI) and functional interference techniques (e.g. transcranial magnetic stimulation: TMS) increasingly address similar questions of functional brain organization, including connectivity, interactions, and causality in the brain. Time-resolved TMS over multiple brain network nodes can elucidate the relative timings of functional relevance for behavior (“TMS chronometry”), while fMRI functional or effective connectivity (fMRI EC) can map task-specific interactions between brain regions based on the interrelation of measured signals. The current study empirically assessed the relation between these different methods.

Methodology/Principal Findings

One group of 15 participants took part in two experiments: one fMRI EC study, and one TMS chronometry study, both of which used an established cognitive paradigm involving one visuospatial judgment task and one color judgment control task. Granger causality mapping (GCM), a data-driven variant of fMRI EC analysis, revealed a frontal-to-parietal flow of information, from inferior/middle frontal gyrus (MFG) to posterior parietal cortex (PPC). FMRI EC-guided Neuronavigated TMS had behavioral effects when applied to both PPC and to MFG, but the temporal pattern of these effects was similar for both stimulation sites. At first glance, this would seem in contradiction to the fMRI EC results. However, we discuss how TMS chronometry and fMRI EC are conceptually different and show how they can be complementary and mutually constraining, rather than contradictory, on the basis of our data.

Conclusions/Significance

The findings that fMRI EC could successfully localize functionally relevant TMS target regions on the single subject level, and conversely, that TMS confirmed an fMRI EC identified functional network to be behaviorally relevant, have important methodological and theoretical implications. Our results, in combination with data from earlier studies by our group (Sack et al., 2007, Cerebral Cortex), lead to informed speculations on complex brain mechanisms, and TMS disruption thereof, underlying visuospatial judgment. This first in-depth empirical and conceptual comparison of fMRI EC and TMS chronometry thereby shows the complementary insights offered by the two methods.     

Dynamic Neural Network of Insight: A Functional Magnetic Resonance Imaging Study on Solving Chinese ‘Chengyu’ Riddles

The key components of insight include breaking mental sets and forming the novel, task-related associations. The majority of researchers have agreed that the anterior cingulate cortex may mediate processes of breaking one’s mental set, while the exact neural correlates of forming novel associations are still debatable. In the present study, we used a paradigm of answer selection to explore brain activations of insight by using event-related functional magnetic resonance imaging during solving Chinese ‘chengyu’ (in Chinese pinyin) riddles. Based on the participant’s choice, the trials were classified into the insight and non-insight conditions. Both stimulus-locked and response-locked analyses are conducted to detect the neural activity corresponding to the early and late periods of insight solution, respectively. Our data indicate that the early period of insight solution shows more activation in the middle temporal gyrus, the middle frontal gyrus and the anterior cingulate cortex. These activities might be associated to the extensive semantic processing, as well as detecting and resolving cognitive conflicts. In contrast, the late period of insight solution produced increased activities in the hippocampus and the amygdala, possibly reflecting the forming of novel association and the concomitant “Aha” feeling. Our study supports the key role of hippocampus in forming novel associations, and indicates a dynamic neural network during insight solution.