Decoding continuous variables from neuroimaging data: basic and clinical applications

Machine learning applied to fMRI patterns of brain activation in response to mutilation pictures predicts PTSD symptoms

BACKGROUND

The present study aimed to apply multivariate pattern recognition methods to predict posttraumatic stress symptoms from whole-brain activation patterns during two contexts where the aversiveness of unpleasant pictures was manipulated by the presence or absence of safety cues.

METHODS

Trauma-exposed participants were presented with neutral and mutilation pictures during functional magnetic resonance imaging (fMRI) collection. Before the presentation of pictures, a text informed the subjects that the pictures were fictitious ("safe context") or real-life scenes ("real context"). We trained machine learning regression models (Gaussian process regression (GPR)) to predict PTSD symptoms in real and safe contexts.

RESULTS

The GPR model could predict PTSD symptoms from brain responses to mutilation pictures in the real context but not in the safe context. The brain regions with the highest contribution to the model were the occipito-parietal regions, including the superior parietal gyrus, inferior parietal gyrus, and supramarginal gyrus. Additional analysis showed that GPR regression models accurately predicted clusters of PTSD symptoms, nominal intrusion, avoidance, and alterations in cognition. As expected, we obtained very similar results as those obtained in a model predicting PTSD total symptoms.

CONCLUSION

This study is the first to show that machine learning applied to fMRI data collected in an aversive context can predict not only PTSD total symptoms but also clusters of PTSD symptoms in a more aversive context. Furthermore, this approach was able to identify potential biomarkers for PTSD, especially in occipitoparietal regions.

Prediction of individual trigeminal pain sensitivity from gray matter structure within the sensorimotor network

OBJECTIVE

To determine whether multivariate pattern regression analysis based on gray matter (GM) images constrained to the sensorimotor network could accurately predict trigeminal heat pain sensitivity in healthy individuals.

BACKGROUND

Prediction of individual pain sensitivity is of clinical relevance as high pain sensitivity is associated with increased risks of postoperative pain, pain chronification, and a poor treatment response. However, as pain is a subjective experience accurate identification of such individuals can be difficult. GM structure of sensorimotor regions have been shown to vary with pain sensitivity. It is unclear whether GM structure within these regions can be used to predict pain sensitivity.

METHODS

In this cross-sectional study, structural magnetic resonance images and pain thresholds in response to contact heat stimulation of the left supraorbital area were obtained from 79 healthy participants. Voxel-based morphometry was used to extract segmented and normalized GM images. These were then constrained to a mask encompassing the functionally defined resting-state sensorimotor network. The masked images and pain thresholds entered a multivariate relevance vector regression analysis for quantitative prediction of the individual pain thresholds. The correspondence between predicted and actual pain thresholds was indexed by the Pearson correlation coefficient (r) and the mean squared error (MSE). The generalizability of the model was assessed by 10-fold and 5-fold cross-validation. Non-parametric permutation tests were used to estimate significance levels.

RESULTS

Trigeminal heat pain sensitivity could be predicted from GM structure within the sensorimotor network with significant accuracy (10-fold: r = 0.53, p < 0.001, MSE = 10.32, p = 0.001; 5-fold: r = 0.46, p = 0.001, MSE = 10.54, p < 0.001). The resulting multivariate weight maps revealed that accurate prediction relied on multiple widespread regions within the sensorimotor network.

CONCLUSION

A multivariate pattern of GM structure within the sensorimotor network could be used to make accurate predictions about trigeminal heat pain sensitivity at the individual level in healthy participants. Widespread regions within the sensorimotor network contributed to the predictive model.

Predicting subclinical psychotic-like experiences on a continuum using machine learning

Previous studies applying machine learning methods to psychosis have primarily been concerned with the binary classification of chronic schizophrenia patients and healthy controls. The aim of this study was to use electroencephalographic (EEG) data and pattern recognition to predict subclinical psychotic-like experiences on a continuum between these two extremes in otherwise healthy people. We applied two different approaches to an auditory oddball regularity learning task obtained from N = 73 participants: A feature extraction and selection routine incorporating behavioural measures, event-related potential components and effective connectivity parameters; Regularisation of spatiotemporal maps of event-related potentials. Using the latter approach, optimal performance was achieved using the response to frequent, predictable sounds. Features within the P50 and P200 time windows had the greatest contribution toward lower Prodromal Questionnaire (PQ) scores and the N100 time window contributed most to higher PQ scores. As a proof-of-concept, these findings demonstrate that EEG data alone are predictive of individual psychotic-like experiences in healthy people. Our findings are in keeping with the mounting evidence for altered sensory responses in schizophrenia, as well as the notion that psychosis may exist on a continuum expanding into the non-clinical population.

In vivo biomarkers of structural and functional brain development and aging in humans

Brain aging is a major determinant of aging. Along with the aging population, prevalence of neurodegenerative diseases is increasing, therewith placing economic and social burden on individuals and society. Individual rates of brain aging are shaped by genetics, epigenetics, and prenatal environmental. Biomarkers of biological brain aging are needed to predict individual trajectories of aging and the risk for age-associated neurological impairments for developing early preventive and interventional measures. We review current advances of in vivo biomarkers predicting individual brain age. Telomere length and epigenetic clock, two important biomarkers that are closely related to the mechanistic aging process, have only poor deterministic and predictive accuracy regarding individual brain aging due to their high intra- and interindividual variability. Phenotype-related biomarkers of global cognitive function and brain structure provide a much closer correlation to age at the individual level. During fetal and perinatal life, autonomic activity is a unique functional marker of brain development. The cognitive and structural biomarkers also boast high diagnostic specificity for determining individual risks for neurodegenerative diseases.

Vulnerability and Protective Factors for PTSD and Depression Symptoms Among Healthcare Workers During COVID-19: A Machine Learning Approach

Background: Healthcare workers are at high risk for developing mental health problems during the COVID-19 pandemic. There is an urgent need to identify vulnerability and protective factors related to the severity of psychiatric symptoms among healthcare workers to implement targeted prevention and intervention programs to reduce the mental health burden worldwide during COVID-19. Objective: The present study aimed to apply a machine learning approach to predict depression and PTSD symptoms based on psychometric questions that assessed: (1) the level of stress due to being isolated from one's family; (2) professional recognition before and during the pandemic; and (3) altruistic acceptance of risk during the COVID-19 pandemic among healthcare workers. Methods: A total of 437 healthcare workers who experienced some level of isolation at the time of the pandemic participated in the study. Data were collected using a web survey conducted between June 12, 2020, and September 19, 2020. We trained two regression models to predict PTSD and depression symptoms. Pattern regression analyses consisted of a linear epsilon-insensitive support vector machine (ε-SVM). Predicted and actual clinical scores were compared using Pearson's correlation coefficient (r), the coefficient of determination (r²), and the normalized mean squared error (NMSE) to evaluate the model performance. A permutation test was applied to estimate significance levels. Results: Results were significant using two different cross-validation strategies to significantly decode both PTSD and depression symptoms. For all of the models, the stress due to social isolation and professional recognition were the variables with the greatest contributions to the predictive function. Interestingly, professional recognition had a negative predictive value, indicating an inverse relationship with PTSD and depression symptoms. Conclusions: Our findings emphasize the protective role of professional recognition and the vulnerability role of the level of stress due to social isolation in the severity of posttraumatic stress and depression symptoms. The insights gleaned from the current study will advance efforts in terms of intervention programs and public health messaging.

Behavior- and Modality-General Representation of Confidence in Orbitofrontal Cortex

Every decision we make is accompanied by a sense of confidence about its likely outcome. This sense informs subsequent behavior, such as investing more-whether time, effort, or money-when reward is more certain. A neural representation of confidence should originate from a statistical computation and predict confidence-guided behavior. An additional requirement for confidence representations to support metacognition is abstraction: they should emerge irrespective of the source of information and inform multiple confidence-guided behaviors. It is unknown whether neural confidence signals meet these criteria. Here, we show that single orbitofrontal cortex neurons in rats encode statistical decision confidence irrespective of the sensory modality, olfactory or auditory, used to make a choice. The activity of these neurons also predicts two confidence-guided behaviors: trial-by-trial time investment and cross-trial choice strategy updating. Orbitofrontal cortex thus represents decision confidence consistent with a metacognitive process that is useful for mediating confidence-guided economic decisions.

Machine Learning Methods Predict Individual Upper-Limb Motor Impairment Following Therapy in Chronic Stroke

Background. Accurate prediction of clinical impairment in upper-extremity motor function following therapy in chronic stroke patients is a difficult task for clinicians but is key in prescribing appropriate therapeutic strategies. Machine learning is a highly promising avenue with which to improve prediction accuracy in clinical practice. Objectives. The objective was to evaluate the performance of 5 machine learning methods in predicting postintervention upper-extremity motor impairment in chronic stroke patients using demographic, clinical, neurophysiological, and imaging input variables. Methods. A total of 102 patients (female: 31%, age 61 ± 11 years) were included. The upper-extremity Fugl-Meyer Assessment (UE-FMA) was used to assess motor impairment of the upper limb before and after intervention. Elastic net (EN), support vector machines, artificial neural networks, classification and regression trees, and random forest were used to predict postintervention UE-FMA. The performances of methods were compared using cross-validated R². Results. EN performed significantly better than other methods in predicting postintervention UE-FMA using demographic and baseline clinical data (median REN2=0.91,RRF2=0.88,RANN2=0.83,RSVM2=0.79,RCART2=0.70; P < .05). Preintervention UE-FMA and the difference in motor threshold (MT) between the affected and unaffected hemispheres were the strongest predictors. The difference in MT had greater importance than the absence or presence of a motor-evoked potential (MEP) in the affected hemisphere. Conclusion. Machine learning methods may enable clinicians to accurately predict a chronic stroke patient's postintervention UE-FMA. Interhemispheric difference in the MT is an important predictor of chronic stroke patients' response to therapy and, therefore, could be included in prospective studies.

Ten Years of BrainAGE as a Neuroimaging Biomarker of Brain Aging: What Insights Have We Gained?

With the aging population, prevalence of neurodegenerative diseases is increasing, thus placing a growing burden on individuals and the whole society. However, individual rates of aging are shaped by a great variety of and the interactions between environmental, genetic, and epigenetic factors. Establishing biomarkers of the neuroanatomical aging processes exemplifies a new trend in neuroscience in order to provide risk-assessments and predictions for age-associated neurodegenerative and neuropsychiatric diseases at a single-subject level. The "Brain Age Gap Estimation (BrainAGE)" method constitutes the first and actually most widely applied concept for predicting and evaluating individual brain age based on structural MRI. This review summarizes all studies published within the last 10 years that have established and utilized the BrainAGE method to evaluate the effects of interaction of genes, environment, life burden, diseases, or life time on individual neuroanatomical aging. In future, BrainAGE and other brain age prediction approaches based on structural or functional markers may improve the assessment of individual risks for neurological, neuropsychiatric and neurodegenerative diseases as well as aid in developing personalized neuroprotective treatments and interventions.

Predicting anxiety from wholebrain activity patterns to emotional faces in young adults: a machine learning approach

BACKGROUND

It is becoming increasingly clear that pathophysiological processes underlying psychiatric disorders categories are heterogeneous on many levels, including symptoms, disease course, comorbidity and biological underpinnings. This heterogeneity poses challenges for identifying biological markers associated with dimensions of symptoms and behaviour that could provide targets to guide treatment choice and novel treatment. In response, the research domain criteria (RDoC) (Insel et al., 2010) was developed to advocate a dimensional approach which omits any disease definitions, disorder thresholds, or cut-points for various levels of psychopathology to understanding the pathophysiological processes underlying psychiatry disorders. In the present study we aimed to apply pattern regression analysis to identify brain signatures during dynamic emotional face processing that are predictive of anxiety and depression symptoms in a continuum that ranges from normal to pathological levels, cutting across categorically-defined diagnoses.

METHODS

The sample was composed of one-hundred and fifty-four young adults (mean age=21.6 and s.d.=2.0, 103 females) consisting of eighty-two young adults seeking treatment for psychological distress that cut across categorically-defined diagnoses and 72 matched healthy young adults. Participants performed a dynamic face task involving fearful, angry and happy faces (and geometric shapes) while undergoing functional Magnetic Resonance Imaging (fMRI). Pattern regression analyses consisted of Gaussian Process Regression (GPR) implemented in the Pattern Recognition for Neuroimaging toolbox (PRoNTo). Predicted and actual clinical scores were compared using Pearson's correlation coefficient (r) and normalized mean squared error (MSE) to evaluate the models' performance. Permutation test was applied to estimate significance levels.

RESULTS

GPR identified patterns of neural activity to dynamic emotional face processing predictive of self-report anxiety in the whole sample, which covered a continuum that ranged from healthy to different levels of distress, including subthreshold to fully-syndromal psychiatric diagnoses. Results were significant using two different cross validation strategies (two-fold: r=0.28 (p-value=0.001), MSE=4.47 (p-value=0.001) and five fold r=0.28 (p-value=0.002), MSE=4.62 (p-value=0.003). The contributions of individual regions to the predictive model were very small, demonstrating that predictions were based on the overall pattern rather than on a small combination of regions.

CONCLUSIONS

These findings represent early evidence that neuroimaging techniques may inform clinical assessment of young adults irrespective of diagnoses by allowing accurate and objective quantitative estimation of psychopathology.

Sparse Ordinal Logistic Regression and Its Application to Brain Decoding

Brain decoding with multivariate classification and regression has provided a powerful framework for characterizing information encoded in population neural activity. Classification and regression models are respectively used to predict discrete and continuous variables of interest. However, cognitive and behavioral parameters that we wish to decode are often ordinal variables whose values are discrete but ordered, such as subjective ratings. To date, there is no established method of predicting ordinal variables in brain decoding. In this study, we present a new algorithm, sparse ordinal logistic regression (SOLR), that combines ordinal logistic regression with Bayesian sparse weight estimation. We found that, in both simulation and analyses using real functional magnetic resonance imaging (fMRI) data, SOLR outperformed ordinal logistic regression with non-sparse regularization, indicating that sparseness leads to better decoding performance. SOLR also outperformed classification and linear regression models with the same type of sparseness, indicating the advantage of the modeling tailored to ordinal outputs. Our results suggest that SOLR provides a principled and effective method of decoding ordinal variables.

Premature brain aging in humans exposed to maternal nutrient restriction during early gestation

BACKGROUND

Prenatal exposure to undernutrition is widespread in both developing and industrialized countries, causing irreversible damage to the developing brain, resulting in altered brain structure and decreased cognitive function during adulthood. The Dutch famine in 1944/45 was a humanitarian disaster, now enabling studies of the effects of prenatal undernutrition during gestation on brain aging in late adulthood.

METHODS

We hypothesized that study participants prenatally exposed to maternal nutrient restriction (MNR) would demonstrate altered brain structure resembling premature brain aging in late adulthood, expecting the effect being stronger in men. Utilizing the Dutch famine birth cohort (n = 118; mean age: 67.5 ± 0.9 years), this study implements an innovative biomarker for individual brain aging, using structural neuroimaging. BrainAGE was calculated using state-of-the-art pattern recognition methods, trained on an independent healthy reference sample, then applied to the Dutch famine MRI sample, to evaluate the effects of prenatal undernutrition during early gestation on individual brain aging in late adulthood.

RESULTS

Exposure to famine in early gestation was associated with BrainAGE scores indicative of an older-appearing brain in the male sample (mean difference to subjects born before famine: 4.3 years, p < 0.05). Furthermore, in explaining the observed variance in individual BrainAGE scores in the male sample, maternal age at birth, head circumference at birth, medical treatment of hypertension, history of cerebral incidences, actual heart rate, and current alcohol intake emerged to be the most influential variables (adjusted R² = 0.63, p < 0.01).

INTERPRETATION

The findings of our study on exposure to prenatal undernutrition being associated with a status of premature brain aging during late adulthood, as well as individual brain structure being shaped by birth- and late-life health characteristics, are strongly supporting the critical importance of sufficient nutrient supply during pregnancy. Interestingly, the status of premature brain aging in participants exposed to the Dutch famine during early gestation occurred in the absence of fetal growth restriction at birth as well as vascular pathology in late-life. Additionally, the neuroimaging brain aging biomarker presented in this study will further enable tracking effects of environmental influences or (preventive) treatments on individual brain maturation and aging in epidemiological and clinical studies.

Mining Time-Resolved Functional Brain Graphs to an EEG-Based Chronnectomic Brain Aged Index (CBAI)

The brain at rest consists of spatially and temporal distributed but functionally connected regions that called intrinsic connectivity networks (ICNs). Resting state electroencephalography (rs-EEG) is a way to characterize brain networks without confounds associated with task EEG such as task difficulty and performance. A novel framework of how to study dynamic functional connectivity under the notion of functional connectivity microstates (FCμstates) and symbolic dynamics is further discussed. Furthermore, we introduced a way to construct a single integrated dynamic functional connectivity graph (IDFCG) that preserves both the strength of the connections between every pair of sensors but also the type of dominant intrinsic coupling modes (DICM). The whole methodology is demonstrated in a significant and unexplored task for EEG which is the definition of an objective Chronnectomic Brain Aged index (CBAI) extracted from resting-state data (N = 94 subjects) with both eyes-open and eyes-closed conditions. Novel features have been defined based on symbolic dynamics and the notion of DICM and FCμstates. The transition rate of FCμstates, the symbolic dynamics based on the evolution of FCμstates (the Markovian Entropy, the complexity index), the probability distribution of DICM, the novel Flexibility Index that captures the dynamic reconfiguration of DICM per pair of EEG sensors and the relative signal power constitute a valuable pool of features that can build the proposed CBAI. Here we applied a feature selection technique and Extreme Learning Machine (ELM) classifier to discriminate young adults from middle-aged and a Support Vector Regressor to build a linear model of the actual age based on EEG-based spatio-temporal features. The most significant type of features for both prediction of age and discrimination of young vs. adults age groups was the dynamic reconfiguration of dominant coupling modes derived from a subset of EEG sensor pairs. Specifically, our results revealed a very high prediction of age for eyes-open (R² = 0.60; y = 0.79x + 8.03) and lower for eyes-closed (R² = 0.48; y = 0.71x + 10.91) while we succeeded to correctly classify young vs. middle-age group with 97.8% accuracy in eyes-open and 87.2% for eyes-closed. Our results were reproduced also in a second dataset for further external validation of the whole analysis. The proposed methodology proved valuable for the characterization of the intrinsic properties of dynamic functional connectivity through the age untangling developmental differences using EEG resting-state recordings.

Evaluating the diagnostic utility of applying a machine learning algorithm to diffusion tensor MRI measures in individuals with major depressive disorder

Using MRI to diagnose mental disorders has been a long-term goal. Despite this, the vast majority of prior neuroimaging work has been descriptive rather than predictive. The current study applies support vector machine (SVM) learning to MRI measures of brain white matter to classify adults with Major Depressive Disorder (MDD) and healthy controls. In a precisely matched group of individuals with MDD (n =25) and healthy controls (n =25), SVM learning accurately (74%) classified patients and controls across a brain map of white matter fractional anisotropy values (FA). The study revealed three main findings: 1) SVM applied to DTI derived FA maps can accurately classify MDD vs. healthy controls; 2) prediction is strongest when only right hemisphere white matter is examined; and 3) removing FA values from a region identified by univariate contrast as significantly different between MDD and healthy controls does not change the SVM accuracy. These results indicate that SVM learning applied to neuroimaging data can classify the presence versus absence of MDD and that predictive information is distributed across brain networks rather than being highly localized. Finally, MDD group differences revealed through typical univariate contrasts do not necessarily reveal patterns that provide accurate predictive information.

Classification and Extraction of Resting State Networks Using Healthy and Epilepsy fMRI Data

Functional magnetic resonance imaging studies have significantly expanded the field's understanding of functional brain activity of healthy and patient populations. Resting state (rs-) fMRI, which does not require subjects to perform a task, eliminating confounds of task difficulty, allows examination of neural activity and offers valuable functional mapping information. The purpose of this work was to develop an automatic resting state network (RSN) labeling method which offers value in clinical workflow during rs-fMRI mapping by organizing and quickly labeling spatial maps into functional networks. Here independent component analysis (ICA) and machine learning were applied to rs-fMRI data with the goal of developing a method for the clinically oriented task of extracting and classifying spatial maps into auditory, visual, default-mode, sensorimotor, and executive control RSNs from 23 epilepsy patients (and for general comparison, separately for 30 healthy subjects). ICA revealed distinct and consistent functional network components across patients and healthy subjects. Network classification was successful, achieving 88% accuracy for epilepsy patients with a naïve Bayes algorithm (and 90% accuracy for healthy subjects with a perceptron). The method's utility to researchers and clinicians is the provided RSN spatial maps and their functional labeling which offer complementary functional information to clinicians' expert interpretation.

Decoding post-stroke motor function from structural brain imaging

Clinical research based on neuroimaging data has benefited from machine learning methods, which have the ability to provide individualized predictions and to account for the interaction among units of information in the brain. Application of machine learning in structural imaging to investigate diseases that involve brain injury presents an additional challenge, especially in conditions like stroke, due to the high variability across patients regarding characteristics of the lesions. Extracting data from anatomical images in a way that translates brain damage information into features to be used as input to learning algorithms is still an open question. One of the most common approaches to capture regional information from brain injury is to obtain the lesion load per region (i.e. the proportion of voxels in anatomical structures that are considered to be damaged). However, no systematic evaluation has yet been performed to compare this approach with using patterns of voxels (i.e. considering each voxel as a single feature). In this paper we compared both approaches applying Gaussian Process Regression to decode motor scores in 50 chronic stroke patients based solely on data derived from structural MRI. For both approaches we compared different ways to delimit anatomical areas: regions of interest from an anatomical atlas, the corticospinal tract, a mask obtained from fMRI analysis with a motor task in healthy controls and regions selected using lesion-symptom mapping. Our analysis showed that extracting features through patterns of voxels that represent lesion probability produced better results than quantifying the lesion load per region. In particular, from the different ways to delimit anatomical areas compared, the best performance was obtained with a combination of a range of cortical and subcortical motor areas as well as the corticospinal tract. These results will inform the appropriate methodology for predicting long term motor outcomes from early post-stroke structural brain imaging.

Evolving Evidence for the Value of Neuroimaging Methods and Biological Markers in Subjects Categorized with Subjective Cognitive Decline

There is evolving evidence that individuals categorized with subjective cognitive decline (SCD) are potentially at higher risk for developing objective and progressive cognitive impairment compared to cognitively healthy individuals without apparent subjective complaints. Interestingly, SCD, during advancing preclinical Alzheimer's disease (AD), may denote very early, subtle cognitive decline that cannot be identified using established standardized tests of cognitive performance. The substantial heterogeneity of existing SCD-related research data has led the Subjective Cognitive Decline Initiative (SCD-I) to accomplish an international consensus on the definition of a conceptual research framework on SCD in preclinical AD. In the area of biological markers, the cerebrospinal fluid signature of AD has been reported to be more prevalent in subjects with SCD compared to healthy controls; moreover, there is a pronounced atrophy, as demonstrated by magnetic resonance imaging, and an increased hypometabolism, as revealed by positron emission tomography, in characteristic brain regions affected by AD. In addition, SCD individuals carrying an apolipoprotein ɛ4 allele are more likely to display AD-phenotypic alterations. The urgent requirement to detect and diagnose AD as early as possible has led to the critical examination of the diagnostic power of biological markers, neurophysiology, and neuroimaging methods for AD-related risk and clinical progression in individuals defined with SCD. Observational studies on the predictive value of SCD for developing AD may potentially be of practical value, and an evidence-based, validated, qualified, and fully operationalized concept may inform clinical diagnostic practice and guide earlier designs in future therapy trials.

Can Emotional and Behavioral Dysregulation in Youth Be Decoded from Functional Neuroimaging?

INTRODUCTION

High comorbidity among pediatric disorders characterized by behavioral and emotional dysregulation poses problems for diagnosis and treatment, and suggests that these disorders may be better conceptualized as dimensions of abnormal behaviors. Furthermore, identifying neuroimaging biomarkers related to dimensional measures of behavior may provide targets to guide individualized treatment. We aimed to use functional neuroimaging and pattern regression techniques to determine whether patterns of brain activity could accurately decode individual-level severity on a dimensional scale measuring behavioural and emotional dysregulation at two different time points.

METHODS

A sample of fifty-seven youth (mean age: 14.5 years; 32 males) was selected from a multi-site study of youth with parent-reported behavioral and emotional dysregulation. Participants performed a block-design reward paradigm during functional Magnetic Resonance Imaging (fMRI). Pattern regression analyses consisted of Relevance Vector Regression (RVR) and two cross-validation strategies implemented in the Pattern Recognition for Neuroimaging toolbox (PRoNTo). Medication was treated as a binary confounding variable. Decoded and actual clinical scores were compared using Pearson's correlation coefficient (r) and mean squared error (MSE) to evaluate the models. Permutation test was applied to estimate significance levels.

RESULTS

Relevance Vector Regression identified patterns of neural activity associated with symptoms of behavioral and emotional dysregulation at the initial study screen and close to the fMRI scanning session. The correlation and the mean squared error between actual and decoded symptoms were significant at the initial study screen and close to the fMRI scanning session. However, after controlling for potential medication effects, results remained significant only for decoding symptoms at the initial study screen. Neural regions with the highest contribution to the pattern regression model included cerebellum, sensory-motor and fronto-limbic areas.

CONCLUSIONS

The combination of pattern regression models and neuroimaging can help to determine the severity of behavioral and emotional dysregulation in youth at different time points.

Decoding negative affect personality trait from patterns of brain activation to threat stimuli

INTRODUCTION

Pattern recognition analysis (PRA) applied to functional magnetic resonance imaging (fMRI) has been used to decode cognitive processes and identify possible biomarkers for mental illness. In the present study, we investigated whether the positive affect (PA) or negative affect (NA) personality traits could be decoded from patterns of brain activation in response to a human threat using a healthy sample.

METHODS

fMRI data from 34 volunteers (15 women) were acquired during a simple motor task while the volunteers viewed a set of threat stimuli that were directed either toward them or away from them and matched neutral pictures. For each participant, contrast images from a General Linear Model (GLM) between the threat versus neutral stimuli defined the spatial patterns used as input to the regression model. We applied a multiple kernel learning (MKL) regression combining information from different brain regions hierarchically in a whole brain model to decode the NA and PA from patterns of brain activation in response to threat stimuli.

RESULTS

The MKL model was able to decode NA but not PA from the contrast images between threat stimuli directed away versus neutral with a significance above chance. The correlation and the mean squared error (MSE) between predicted and actual NA were 0.52 (p-value=0.01) and 24.43 (p-value=0.01), respectively. The MKL pattern regression model identified a network with 37 regions that contributed to the predictions. Some of the regions were related to perception (e.g., occipital and temporal regions) while others were related to emotional evaluation (e.g., caudate and prefrontal regions).

CONCLUSION

These results suggest that there was an interaction between the individuals' NA and the brain response to the threat stimuli directed away, which enabled the MKL model to decode NA from the brain patterns. To our knowledge, this is the first evidence that PRA can be used to decode a personality trait from patterns of brain activation during emotional contexts.

Cross-validation and hypothesis testing in neuroimaging: An irenic comment on the exchange between Friston and Lindquist et al

The "ten ironic rules for statistical reviewers" presented by Friston (2012) prompted a rebuttal by Lindquist et al. (2013), which was followed by a rejoinder by Friston (2013). A key issue left unresolved in this discussion is the use of cross-validation to test the significance of predictive analyses. This note discusses the role that cross-validation-based and related hypothesis tests have come to play in modern data analyses, in neuroimaging and other fields. It is shown that such tests need not be suboptimal and can fill otherwise-unmet inferential needs.

Individual differences in visual field shape modulate the effects of attention on the lower visual field advantage in crowding

It has previously been reported that visual crowding of a target by flankers is stronger in the upper visual field than in the lower, and this finding has been attributed to greater attentional resolution in the lower hemifield (He, Cavanagh, & Intriligator, 1996). Here we show that the upper/lower asymmetry in visual crowding can be explained by natural variations in the borders of each individual's visual field. Specifically, asymmetry in crowding along the vertical meridian can be almost entirely accounted for by replacing the conventional definition of visual field location, in units of degrees of visual angle, with a definition based on the ratio of the extents of an individual's upper and lower visual field. We also show that the upper/lower crowding asymmetry is eliminated when stimulus eccentricity is expressed in units of percentage of visual field extent but is present when the conventional measure of visual angle is used. We further demonstrate that the relationship between visual field extent and perceptual asymmetry is most evident when participants are able to focus their attention on the target location. These results reveal important influences of visual field boundaries on visual perception, even for visual field locations far from those boundaries.

Disease prediction based on functional connectomes using a scalable and spatially-informed support vector machine

Substantial evidence indicates that major psychiatric disorders are associated with distributed neural dysconnectivity, leading to a strong interest in using neuroimaging methods to accurately predict disorder status. In this work, we are specifically interested in a multivariate approach that uses features derived from whole-brain resting state functional connectomes. However, functional connectomes reside in a high dimensional space, which complicates model interpretation and introduces numerous statistical and computational challenges. Traditional feature selection techniques are used to reduce data dimensionality, but are blind to the spatial structure of the connectomes. We propose a regularization framework where the 6-D structure of the functional connectome (defined by pairs of points in 3-D space) is explicitly taken into account via the fused Lasso or the GraphNet regularizer. Our method only restricts the loss function to be convex and margin-based, allowing non-differentiable loss functions such as the hinge-loss to be used. Using the fused Lasso or GraphNet regularizer with the hinge-loss leads to a structured sparse support vector machine (SVM) with embedded feature selection. We introduce a novel efficient optimization algorithm based on the augmented Lagrangian and the classical alternating direction method, which can solve both fused Lasso and GraphNet regularized SVM with very little modification. We also demonstrate that the inner subproblems of the algorithm can be solved efficiently in analytic form by coupling the variable splitting strategy with a data augmentation scheme. Experiments on simulated data and resting state scans from a large schizophrenia dataset show that our proposed approach can identify predictive regions that are spatially contiguous in the 6-D "connectome space," offering an additional layer of interpretability that could provide new insights about various disease processes.

Quantifying anatomical shape variations in neurological disorders

We develop a multivariate analysis of brain anatomy to identify the relevant shape deformation patterns and quantify the shape changes that explain corresponding variations in clinical neuropsychological measures. We use kernel Partial Least Squares (PLS) and formulate a regression model in the tangent space of the manifold of diffeomorphisms characterized by deformation momenta. The scalar deformation momenta completely encode the diffeomorphic changes in anatomical shape. In this model, the clinical measures are the response variables, while the anatomical variability is treated as the independent variable. To better understand the "shape-clinical response" relationship, we also control for demographic confounders, such as age, gender, and years of education in our regression model. We evaluate the proposed methodology on the Alzheimer's Disease Neuroimaging Initiative (ADNI) database using baseline structural MR imaging data and neuropsychological evaluation test scores. We demonstrate the ability of our model to quantify the anatomical deformations in units of clinical response. Our results also demonstrate that the proposed method is generic and generates reliable shape deformations both in terms of the extracted patterns and the amount of shape changes. We found that while the hippocampus and amygdala emerge as mainly responsible for changes in test scores for global measures of dementia and memory function, they are not a determinant factor for executive function. Another critical finding was the appearance of thalamus and putamen as most important regions that relate to executive function. These resulting anatomical regions were consistent with very high confidence irrespective of the size of the population used in the study. This data-driven global analysis of brain anatomy was able to reach similar conclusions as other studies in Alzheimer's disease based on predefined ROIs, together with the identification of other new patterns of deformation. The proposed methodology thus holds promise for discovering new patterns of shape changes in the human brain that could add to our understanding of disease progression in neurological disorders.

Decoding the neuroanatomical basis of reading ability: a multivoxel morphometric study

As a relatively recent cultural invention in human evolution, reading is an important gateway to personal development and socioeconomic success. Despite the well documented individual differences in reading ability, its neuroanatomical correlates have not been well understood, largely due to the fact that reading is a complex skill that consists of multiple components. Using a large sample of 416 college students and 7 reading tasks, the present study successfully identified three uncorrelated components of reading ability: phonological decoding, form-sound association, and naming speed. We then tried to predict individuals' scores in these components from their gray matter volume (GMV) on a subset of participants (N = 253) with high-quality structural images, adopting a multivariate support vector regression analysis with tenfold cross-validation. Our results revealed distinct neural regions that supported different aspects of reading ability: whereas phonological decoding was associated with the GMV in the left superior parietal lobe extending to the supramarginal gyrus, form-sound association was predicted by the GMV in the hippocampus and cerebellum. Naming speed was associated with GMV in distributed brain regions in the occipital, temporal, parietal, and frontal cortices. Phonological decoding and form-sound association were uncorrelated with general cognitive abilities. However, naming speed was correlated with intelligence and processing speed, and some of the regions that were predictive of naming speed also predicted these general cognitive abilities. These results provide further insights on the cognitive and neural architecture of reading and the structural basis of individual differences in reading abilities.

A Generalized ideal observer model for decoding sensory neural responses

We show that many ideal observer models used to decode neural activity can be generalized to a conceptually and analytically simple form. This enables us to study the statistical properties of this class of ideal observer models in a unified manner. We consider in detail the problem of estimating the performance of this class of models. We formulate the problem de novo by deriving two equivalent expressions for the performance and introducing the corresponding estimators. We obtain a lower bound on the number of observations (N) required for the estimate of the model performance to lie within a specified confidence interval at a specified confidence level. We show that these estimators are unbiased and consistent, with variance approaching zero at the rate of 1/N. We find that the maximum likelihood estimator for the model performance is not guaranteed to be the minimum variance estimator even for some simple parametric forms (e.g., exponential) of the underlying probability distributions. We discuss the application of these results for designing and interpreting neurophysiological experiments that employ specific instances of this ideal observer model.

Multivariate linear regression of high-dimensional fMRI data with multiple target variables

Multivariate regression is increasingly used to study the relation between fMRI spatial activation patterns and experimental stimuli or behavioral ratings. With linear models, informative brain locations are identified by mapping the model coefficients. This is a central aspect in neuroimaging, as it provides the sought-after link between the activity of neuronal populations and subject's perception, cognition or behavior. Here, we show that mapping of informative brain locations using multivariate linear regression (MLR) may lead to incorrect conclusions and interpretations. MLR algorithms for high dimensional data are designed to deal with targets (stimuli or behavioral ratings, in fMRI) separately, and the predictive map of a model integrates information deriving from both neural activity patterns and experimental design. Not accounting explicitly for the presence of other targets whose associated activity spatially overlaps with the one of interest may lead to predictive maps of troublesome interpretation. We propose a new model that can correctly identify the spatial patterns associated with a target while achieving good generalization. For each target, the training is based on an augmented dataset, which includes all remaining targets. The estimation on such datasets produces both maps and interaction coefficients, which are then used to generalize. The proposed formulation is independent of the regression algorithm employed. We validate this model on simulated fMRI data and on a publicly available dataset. Results indicate that our method achieves high spatial sensitivity and good generalization and that it helps disentangle specific neural effects from interaction with predictive maps associated with other targets.

Bayesian scalar-on-image regression with application to association between intracranial DTI and cognitive outcomes

Diffusion tensor imaging (DTI) measures water diffusion within white matter, allowing for in vivo quantification of brain pathways. These pathways often subserve specific functions, and impairment of those functions is often associated with imaging abnormalities. As a method for predicting clinical disability from DTI images, we propose a hierarchical Bayesian "scalar-on-image" regression procedure. Our procedure introduces a latent binary map that estimates the locations of predictive voxels and penalizes the magnitude of effect sizes in these voxels, thereby resolving the ill-posed nature of the problem. By inducing a spatial prior structure, the procedure yields a sparse association map that also maintains spatial continuity of predictive regions. The method is demonstrated on a simulation study and on a study of association between fractional anisotropy and cognitive disability in a cross-sectional sample of 135 multiple sclerosis patients.

Bipolar disorder diagnosis: challenges and future directions

Bipolar disorder refers to a group of affective disorders, which together are characterised by depressive and manic or hypomanic episodes. These disorders include: bipolar disorder type I (depressive and manic episodes: this disorder can be diagnosed on the basis of one manic episode); bipolar disorder type II (depressive and hypomanic episodes); cyclothymic disorder (hypomanic and depressive symptoms that do not meet criteria for depressive episodes); and bipolar disorder not otherwise specified (depressive and hypomanic-like symptoms that do not meet the diagnostic criteria for any of the aforementioned disorders). Bipolar disorder type II is especially difficult to diagnose accurately because of the difficulty in differentiation of this disorder from recurrent unipolar depression (recurrent depressive episodes) in depressed patients. The identification of objective biomarkers that represent pathophysiologic processes that differ between bipolar disorder and unipolar depression can both inform bipolar disorder diagnosis and provide biological targets for the development of new and personalised treatments. Neuroimaging studies could help the identification of biomarkers that differentiate bipolar disorder from unipolar depression, but the problem in detection of a clear boundary between these disorders suggests that they might be better represented as a continuum of affective disorders. Innovative combinations of neuroimaging and pattern recognition approaches can identify individual patterns of neural structure and function that accurately ascertain where a patient might lie on a behavioural scale. Ultimately, an integrative approach, with several biological measurements using different scales, could yield patterns of biomarkers (biosignatures) to help identify biological targets for personalised and new treatments for all affective disorders.

Characterizing Functional Connectivity Differences in Aging Adults using Machine Learning on Resting State fMRI Data

The brain at rest consists of spatially distributed but functionally connected regions, called intrinsic connectivity networks (ICNs). Resting state functional magnetic resonance imaging (rs-fMRI) has emerged as a way to characterize brain networks without confounds associated with task fMRI such as task difficulty and performance. Here we applied a Support Vector Machine (SVM) linear classifier as well as a support vector machine regressor to rs-fMRI data in order to compare age-related differences in four of the major functional brain networks: the default, cingulo-opercular, fronto-parietal, and sensorimotor. A linear SVM classifier discriminated between young and old subjects with 84% accuracy (p-value < 1 × 10⁻⁷). A linear SVR age predictor performed reasonably well in continuous age prediction (R² = 0.419, p-value < 1 × 10⁻⁸). These findings reveal that differences in intrinsic connectivity as measured with rs-fMRI exist between subjects, and that SVM methods are capable of detecting and utilizing these differences for classification and prediction.

Decoding the anatomical network of spatial attention

The study of stroke patients with modern lesion-symptom analysis techniques has yielded valuable insights into the representation of spatial attention in the human brain. Here we introduce an approach--multivariate pattern analysis--that no longer assumes independent contributions of brain regions but rather quantifies the joint contribution of multiple brain regions in determining behavior. In a large sample of stroke patients, we found patterns of damage more predictive of spatial neglect than the best-performing single voxel. In addition, modeling multiple brain regions--those that are frequently damaged and, importantly, spared--provided more predictive information than modeling single regions. Interestingly, we also found that the superior temporal gyrus demonstrated a consistent ability to improve classifier performance when added to other regions, implying uniquely predictive information. In sharp contrast, classifier performance for both the angular gyrus and insular cortex was reliably enhanced by the addition of other brain regions, suggesting these regions lack independent predictive information for spatial neglect. Our findings highlight the utility of multivariate pattern analysis in lesion mapping, furnishing neuroscience with a modern approach for using lesion data to study human brain function.

The clinical value of large neuroimaging data sets in Alzheimer's disease

Rapid advances in neuroimaging and cyberinfrastructure technologies have brought explosive growth in the Web-based warehousing, availability, and accessibility of imaging data on a variety of neurodegenerative and neuropsychiatric disorders and conditions. There has been a prolific development and emergence of complex computational infrastructures that serve as repositories of databases and provide critical functionalities such as sophisticated image analysis algorithm pipelines and powerful three-dimensional visualization and statistical tools. The statistical and operational advantages of collaborative, distributed team science in the form of multisite consortia push this approach in a diverse range of population-based investigations.