Functional Brain Imaging Data Research Articles

Should we use the computational or the network approach to analyze functional brain-imaging data—why not both?

Should we use the computational or the network approach to analyze functional brain-imaging data—why not both?

A hybrid spatiotemporal deep belief network and sparse representation-based framework reveals multilevel core functional components in decoding multitask fMRI signals.

Decoding human brain activity on various task-based functional brain imaging data is of great significance for uncovering the functioning mechanism of the human mind. Currently, most feature extraction model-based methods for brain state decoding are shallow machine learning models, which may struggle to capture complex and precise spatiotemporal patterns of brain activity from the highly noisy fMRI raw data. Moreover, although decoding models based on deep learning methods benefit from their multilayer structure that could extract spatiotemporal features at multiscale, the relatively large populations of fMRI datasets are indispensable, and the explainability of their results is elusive. To address the above problems, we proposed a computational framework based on hybrid spatiotemporal deep belief network and sparse representations to differentiate multitask fMRI (tfMRI) signals. Using a relatively small cohort of tfMRI data as a test bed, our framework can achieve an average classification accuracy of 97.86% and define the multilevel temporal and spatial patterns of multiple cognitive tasks. Intriguingly, our model can characterize the key components for differentiating the multitask fMRI signals. Overall, the proposed framework can identify the interpretable and discriminative fMRI composition patterns at multiple scales, offering an effective methodology for basic neuroscience and clinical research with relatively small cohorts.

Individual differences in the language task-evoked and resting-state functional networks.

The resting state functional network is highly variable across individuals. However, inter-individual differences in functional networks evoked by language tasks and their comparison with resting state are still unclear. To address these two questions, we used T1 anatomical data and functional brain imaging data of resting state and a story comprehension task from the Human Connectome Project (HCP) to characterize functional network variability and investigate the uniqueness of the functional network in both task and resting states. We first demonstrated that intrinsic and task-induced functional networks exhibited remarkable differences across individuals, and language tasks can constrain inter-individual variability in the functional brain network. Furthermore, we found that the inter-individual variability of functional networks in two states was broadly consistent and spatially heterogeneous, with high-level association areas manifesting more significant variability than primary visual processing areas. Our results suggested that the functional network underlying language comprehension is unique at the individual level, and the inter-individual variability architecture of the functional network is broadly consistent in language task and resting state.

Diagnosis of autism spectrum disorder based on functional brain networks and machine learning

Autism is a multifaceted neurodevelopmental condition whose accurate diagnosis may be challenging because the associated symptoms and severity vary considerably. The wrong diagnosis can affect families and the educational system, raising the risk of depression, eating disorders, and self-harm. Recently, many works have proposed new methods for the diagnosis of autism based on machine learning and brain data. However, these works focus on only one pairwise statistical metric, ignoring the brain network organization. In this paper, we propose a method for the automatic diagnosis of autism based on functional brain imaging data recorded from 500 subjects, where 242 present autism spectrum disorder considering the regions of interest throughout Bootstrap Analysis of Stable Cluster map. Our method can distinguish the control group from autism spectrum disorder patients with high accuracy. Indeed the best performance provides an AUC near 1.0, which is higher than that found in the literature. We verify that the left ventral posterior cingulate cortex region is less connected to an area in the cerebellum of patients with this neurodevelopment disorder, which agrees with previous studies. The functional brain networks of autism spectrum disorder patients show more segregation, less distribution of information across the network, and less connectivity compared to the control cases. Our workflow provides medical interpretability and can be used on other fMRI and EEG data, including small data sets.

Association between Long-Term Air Pollution, Chronic Traffic Noise, and Resting-State Functional Connectivity in the 1000BRAINS Study.

Background:Older adults show a high variability in cognitive performance that cannot be explained by aging alone. Although research has linked air pollution and noise to cognitive impairment and structural brain alterations, the potential impact of air pollution and noise on functional brain organization is unknown.Objective:This study examined the associations between long-term air pollution and traffic noise with measures of functional brain organization in older adults. We hypothesize that exposures to high air pollution and noise levels are associated with age-like changes in functional brain organization, shown by less segregated brain networks.Methods:Data from 574 participants (44.1% female, 56–85 years of age) in the German 1000BRAINS study (2011–2015) were analyzed. Exposure to particulate matter (, , and absorbance), accumulation mode particle number (), and nitrogen dioxide () was estimated applying land-use regression and chemistry transport models. Noise exposures were assessed as weighted 24-h () and nighttime () means. Functional brain organization of seven established brain networks (visual, sensorimotor, dorsal and ventral attention, limbic, frontoparietal and default network) was assessed using resting-state functional brain imaging data. To assess functional brain organization, we determined the degree of segregation between networks by comparing the strength of functional connections within and between networks. We estimated associations between air pollution and noise exposure with network segregation, applying multiple linear regression models adjusted for age, sex, socioeconomic status, and lifestyle variables.Results:Overall, small associations of high exposures with lesser segregated networks were visible. For the sensorimotor networks, we observed small associations between high air pollution and noise and lower network segregation, which had a similar effect size as a 1-y increase in age [e.g., in sensorimotor network, (95% CI: , 0.009) per 0.3 increase in absorbance and (95% CI: , ) per 1-y age increase].Conclusion:High exposure to air pollution and noise was associated with less segregated functional brain networks. https://doi.org/10.1289/EHP9737

Digitising brain age

Chronological age can differ from biological age, particularly in Alzheimer's disease (AD) and related neurodegenerative disorders. By training models with structural and functional brain imaging data obtained serially over time, deep learning artificial intelligence can estimate the brain age gap—the difference between chronological age and predicted age.

Impaired Arithmetic Fact Retrieval in an Adult with Developmental Dyscalculia: Evidence from Behavioral and Functional Brain Imaging Data.

Developmental dyscalculia (DD) is a developmental disorder characterized by arithmetic difficulties. Recently, it has been suggested that the neural networks supporting procedure-based calculation (e.g., in subtraction) and left-hemispheric verbal arithmetic fact retrieval (e.g., in multiplication) are partially distinct. Here we compared the neurofunctional correlates of subtraction and multiplication in a 19-year-old student (RM) with DD to 18 age-matched controls. Behaviorally, RM performed significantly worse than controls in multiplication, while subtraction was unaffected. Neurofunctional differences were most pronounced regarding multiplication: RM showed significantly stronger activation than controls not only in left angular gyrus but also in a fronto-parietal network (including left intraparietal sulcus and inferior frontal gyrus) typically activated during procedure-based calculation. Region-of-interest analyses indicated group differences in multiplication only, which, however, did not survive correction for multiple comparisons. Our results are consistent with dissociable and processing-specific, but not operation-specific neurofunctional networks. Procedure-based calculation is not only associated with subtraction but also with (untrained) multiplication facts. Only after rote learning, facts can be retrieved quasi automatically from memory. We suggest that this learning process and the associated shift in activation patterns has not fully occurred in RM, as reflected in her need to resort to procedure-based strategies to solve multiplication facts.

Dissociating the Neural Correlates of Subjective Visibility from Those of Decision Confidence.

A key goal of consciousness science is identifying neural signatures of being aware versus unaware of simple stimuli. This is often investigated in the context of near-threshold detection, with reports of stimulus awareness being linked to heightened activation in a frontoparietal network. However, because of reports of stimulus presence typically being associated with higher confidence than reports of stimulus absence, these results could be explained by frontoparietal regions encoding stimulus visibility, decision confidence, or both. In an exploratory analysis, we leverage fMRI data from 35 human participants (20 females) to disentangle these possibilities. We first show that, whereas stimulus identity was best decoded from the visual cortex, stimulus visibility (presence vs absence) was best decoded from prefrontal regions. To control for effects of confidence, we then selectively sampled trials before decoding to equalize confidence distributions between absence and presence responses. This analysis revealed striking differences in the neural correlates of subjective visibility in PFC ROIs, depending on whether or not differences in confidence were controlled for. We interpret our findings as highlighting the importance of controlling for metacognitive aspects of the decision process in the search for neural correlates of visual awareness.SIGNIFICANCE STATEMENT While much has been learned over the past two decades about the neural basis of visual awareness, the role of the PFC remains a topic of debate. By applying decoding analyses to functional brain imaging data, we show that prefrontal representations of subjective visibility are contaminated by neural correlates of decision confidence. We propose a new analysis method to control for these metacognitive aspects of awareness reports, and use it to reveal confidence-independent correlates of perceptual judgments in a subset of prefrontal areas.

Multimodal Brain Age Prediction with Feature Selection and Comparison.

Brain age, an estimated biological age from anatomical and/or functional brain imaging data, and its deviation from the chronological age (brain age gap) have shown the potential to serve as biomarkers for characterizing typical brain development, the abnormal aging process, and early indicators of clinical neuropsychiatric problems. In this study, we leverage multimodal brain imaging data for brain age prediction. We studied and compared the performance of individual data modalities (gray matter density in components and regions of interest, cortical and subcortical anatomical features, resting-state functional connectivity) and different combinations of multiple data modalities using data collected from 1417 participants with age between 8 and 22 years. The result indicates that feature selection and multimodal imaging data can improve brain age prediction with linear support vector and partial least squares regression models. We have achieved a mean absolute error of 1.22 years on the test data with 188 features selected equally from all data sources, better than any individual source. After bias correction, the brain age gap was significantly associated with attention accuracy/speed and motor speed in addition to age. Our results conclude that traditional machine learning with proper feature selection can achieve similar if not better performance compared to complex deep learning neural network methods for the used sample size.

Visual Explanation for Identification of the Brain Bases for Developmental Dyslexia on fMRI Data.

Problem: Brain imaging studies of mental health and neurodevelopmental disorders have recently included machine learning approaches to identify patients based solely on their brain activation. The goal is to identify brain-related features that generalize from smaller samples of data to larger ones; in the case of neurodevelopmental disorders, finding these patterns can help understand differences in brain function and development that underpin early signs of risk for developmental dyslexia. The success of machine learning classification algorithms on neurofunctional data has been limited to typically homogeneous data sets of few dozens of participants. More recently, larger brain imaging data sets have allowed for deep learning techniques to classify brain states and clinical groups solely from neurofunctional features. Indeed, deep learning techniques can provide helpful tools for classification in healthcare applications, including classification of structural 3D brain images. The adoption of deep learning approaches allows for incremental improvements in classification performance of larger functional brain imaging data sets, but still lacks diagnostic insights about the underlying brain mechanisms associated with disorders; moreover, a related challenge involves providing more clinically-relevant explanations from the neural features that inform classification.Methods: We target this challenge by leveraging two network visualization techniques in convolutional neural network layers responsible for learning high-level features. Using such techniques, we are able to provide meaningful images for expert-backed insights into the condition being classified. We address this challenge using a dataset that includes children diagnosed with developmental dyslexia, and typical reader children.Results: Our results show accurate classification of developmental dyslexia (94.8%) from the brain imaging alone, while providing automatic visualizations of the features involved that match contemporary neuroscientific knowledge (brain regions involved in the reading process for the dyslexic reader group and brain regions associated with strategic control and attention processes for the typical reader group).Conclusions: Our visual explanations of deep learning models turn the accurate yet opaque conclusions from the models into evidence to the condition being studied.

Illusory Body Ownership Affects the Cortical Response to Vicarious Somatosensation

Fundamental human feelings such as body ownership (“this” body is “my” body) and vicariousness (first-person-like experience of events occurring to others) are based on multisensory integration. Behavioral links between body ownership and vicariousness have been shown, but the neural underpinnings remain largely unexplored. To fill this gap, we investigated the neural effects of altered body ownership on vicarious somatosensation. While recording functional brain imaging data, first, we altered participants’ body ownership by robotically delivering tactile stimulations (“tactile” stroking) in synchrony or not with videos of a virtual hand being brushed (“visual” stroking). Then, we manipulated vicarious somatosensation by showing videos of the virtual hand being touched by a syringe’s plunger (touch) or needle (pain). Only after the alteration of body ownership (synchronous visuo-tactile stroking) and specifically during late epochs of vicarious somatosensation, vicarious pain was associated with lower activation in premotor and anterior cingulate cortices with respect to vicarious touch. At the methodological level, the present study highlights the importance of the neural response’s temporal evolution. At the theoretical level, it shows that the higher-level (cognitive) impact of a lower-level (sensory) body-related processing (visuo-tactile) is not limited to body ownership but also extends to other psychological body-related domains, such as vicarious somatosensation.

Cognitive emotion regulation withstands the stress test: An fMRI study on the effect of acute stress on distraction and reappraisal

Cognitive emotion regulation is a key mechanism for the maintenance of mental health, but may fail, when individuals are exposed to acute stress. To date, it is not well understood whether and to what extent acute stress effects contribute to impairments in emotion regulation capacities as the sparse existing studies have yielded heterogeneous results, indicating that stress timing might be a crucial factor.In the present study, 81 healthy participants underwent either an acute stress task (ScanSTRESS-C; n = 40) or a control condition (n = 41) while lying in the MRI scanner. In the subsequent Cognitive Emotion Regulation Task (CERT), participants were confronted with neutral or negative pictures and instructed to either view them, or regulate their upcoming emotions using either distraction or reappraisal. Subjective ratings of affective state as well as functional brain imaging data served to indicate emotion regulation.The results showed a successful stress manipulation as indicated by group differences in subjective wellbeing, saliva cortisol concentrations, heart rate, and functional brain activity in regions implicated in stress processing. With respect to emotion regulation, CERT data revealed a significant regulation effect at the neural and behavioral level (less negative emotional ratings after reappraisal and distraction trials compared to view trials) in both groups. However, no significant group differences were observed, neither in BOLD responses to the CERT, nor in behavioral ratings.Contrary to previous studies, our study did not reveal further evidence of stress-related effects on emotion regulation, potentially being related to differences between studies in experimental setting, timing, and procedures. This study therefore underlines the need of future studies that disentangle the complex interplay of stress and emotion regulation and identify different factors influencing their bidirectional relationship.

Dynamic Brain Graph Analysis Show Unique Path Trajectory Weights in Schizophrenia

Brain graph analysis of control and patient groups can be a powerful tool for characterizing functional brain imaging data. Most studies do not focus on connected paths that can occur. In this study we examine whether differences can be found in terms of the characteristics of each path, despite the existence of a path between two distinct brain components in control and patient groups.

Parental Education and Left Lateral Orbitofrontal Cortical Activity during N-Back Task: An fMRI Study of American Adolescents.

Introduction. The Orbitofrontal Cortex (OFC) is a cortical structure that has implications in cognition, memory, reward anticipation, outcome evaluation, decision making, and learning. As such, OFC activity correlates with these cognitive brain abilities. Despite research suggesting race and socioeconomic status (SES) indicators such as parental education may be associated with OFC activity, limited knowledge exists on multiplicative effects of race and parental education on OFC activity and associated cognitive ability. Purpose. Using functional brain imaging data from the Adolescent Brain Cognitive Development (ABCD) study, we tested the multiplicative effects of race and parental education on left lateral OFC activity during an N-Back task. In our study, we used a sociological rather than biological theory that conceptualizes race and SES as proxies of access to the opportunity structure and exposure to social adversities rather than innate and non-modifiable brain differences. We explored racial variation in the effect of parental educational attainment, a primary indicator of SES, on left lateral OFC activity during an N-Back task between Black and White 9–10 years old adolescents. Methods. The ABCD study is a national, landmark, multi-center brain imaging investigation of American adolescents. The total sample was 4290 9–10 years old Black or White adolescents. The independent variables were SES indicators, namely family income, parental education, and neighborhood income. The primary outcome was the average beta weight for N-Back (2 back versus 0 back contrast) in ASEG ROI left OFC activity, measured by functional Magnetic Resonance Imaging (fMRI) during an N-Back task. Ethnicity, age, sex, subjective SES, and family structure were the study covariates. For data analysis, we used linear regression models. Results. In White but not Black adolescents, parental education was associated with higher left lateral OFC activity during the N-Back task. In the pooled sample, we found a significant interaction between race and parental education on the outcome, suggesting that high parental education is associated with a larger increase in left OFC activity of White than Black adolescents. Conclusions. For American adolescents, race and SES jointly influence left lateral OFC activity correlated with cognition, memory, decision making, and learning. Given the central role of left lateral OFC activity in learning and memory, our finding calls for additional research on contextual factors that reduce the gain of SES for Black adolescents. Cognitive inequalities are not merely due to the additive effects of race and SES but also its multiplicative effects.

Why do we need an Asian brain template in clinical trials of Alzheimer's disease? A comparative study of brain structures between Asian vs Caucasian brains

AbstractBackgroundThere is a recognized need for the creation of ethnic‐specific brain templates of cognitively normal older adults for the purpose of evaluating structural and functional brain imaging data in the normal aging process and neurodegenerative disease including Alzheimer’s disease. However, previous studies of racial difference in cortical thickness of cognitively normal older adults have been restricted to small sample size and have not adequately controlled biases. We aimed to compare cerebral cortical thickness between cognitively normal older adults of East Asians and of Caucasians adjusting for age, gender, education.MethodWe evaluated the brain cortical thickness by using FreeSurfer on standard T1‐weighted images of cognitively normal older adults in East Asians [n= 321 (60’s = 166, 70’s = 155)] and Caucasians [n= 282 (60’s = 111, 70’s = 171)], and examined the difference in cortical thickness between two races by whole brain analysis (ANCOVA adjusting age, education, MMSE scores, corrected p < 0.05, FDR correction).ResultEast Asians were shown to have a higher cortical thickness in bilateral superior frontal, lateral and medial orbitofrontal, middle temporal, left superior parietal, right inferior parietal, and inferior temporal cortex and lower cortical thickness in bilateral parahippocampus, postcentral, lateral occipital, left transverse temporal, lingual, precuneus, right superior temporal, fusiform cortex than Caucasians. Additionally, interactions between age and race had an impact on cortical thickness in right superior frontal cortex.ConclusionThe findings indicated that there was a significant difference in cortical thickness between cognitively healthy elderly in East Asians and in Caucasians according to the brain regions. The findings could contribute to a better understanding of the race‐specific structural change in the normal aging and provide an importance of an ethnic‐specific template for evaluating the structural and functional brain imaging in the healthy aging and neurodegenerative disease including Alzheimer’s disease.

Neighborhood Poverty and Amygdala Response to Negative Face.

Considerable research has established a link between socioeconomic status (SES) and brain function. While studies have shown a link between poverty status and amygdala response to negative stimuli, a paucity of knowledge exists on whether neighborhood poverty is also independently associated with amygdala hyperactive response to negative stimuli. Using functional brain imaging data, this study tested the association between neighborhood SES and the amygdala's response to negative stimuli. Considering race as a sociological rather than a biological construct, we also explored racial heterogeneity in this association between non-Hispanic Black and non-Hispanic White youth. We borrowed the functional Magnetic Resonance Imaging (fMRI) data of the Adolescent Brain Cognitive Development (ABCD) study. The sample was 2,490 nine to ten year old non-Hispanic Black and non-Hispanic White adolescents. The independent variable was neighborhood income which was treated as a continuous measure. The primary outcomes were the right and left amygdala response to negative face during an N-Back task. Age, sex, race, marital status, and family SES were the covariates. To analyze the data, we used linear regression models. Low neighborhood income was independently associated with a higher level of amygdala response to negative face. Similar results were seen for the right and left amygdala. These effects were significant net of race, age, sex, marital status, and family SES. An association between low neighborhood SES and higher left but not right amygdala response to negative face could be observed for non-Hispanic Black youth. No association between neighborhood SES and left or right amygdala response to negative face could be observed for non-Hispanic White youth. For American youth, particularly non-Hispanic Black youth, living in a poor neighborhood predicts the left amygdala reaction to negative face. This result suggested that Black youth who live in poor neighborhoods are at a high risk of poor emotion regulation. This finding has implications for policy making to reduce inequalities in undesired behavioral and emotional outcomes. Policy solutions to health inequalities should address inequalities in neighborhood SES.

Breaking down the cocktail party: Attentional modulation of cerebral audiovisual speech processing

Recent studies utilizing electrophysiological speech envelope reconstruction have sparked renewed interest in the cocktail party effect by showing that auditory neurons entrain to selectively attended speech. Yet, the neural networks of attention to speech in naturalistic audiovisual settings with multiple sound sources remain poorly understood. We collected functional brain imaging data while participants viewed audiovisual video clips of lifelike dialogues with concurrent distracting speech in the background. Dialogues were presented in a full-factorial design, comprising task (listen to the dialogues vs. ignore them), audiovisual quality and semantic predictability. We used univariate analyses in combination with multivariate pattern analysis (MVPA) to study modulations of brain activity related to attentive processing of audiovisual speech. We found attentive speech processing to cause distinct spatiotemporal modulation profiles in distributed cortical areas including sensory and frontal-control networks. Semantic coherence modulated attention-related activation patterns in the earliest stages of auditory cortical processing, suggesting that the auditory cortex is involved in high-level speech processing. Our results corroborate views that emphasize the dynamic nature of attention, with task-specificity and context as cornerstones of the underlying neuro-cognitive mechanisms.

Identifying Autism Spectrum Disorder From Resting-State fMRI Using Deep Belief Network.

With the increasing prevalence of autism spectrum disorder (ASD), it is important to identify ASD patients for effective treatment and intervention, especially in early childhood. Neuroimaging techniques have been used to characterize the complex biomarkers based on the functional connectivity anomalies in the ASD. However, the diagnosis of ASD still adopts the symptom-based criteria by clinical observation. The existing computational models tend to achieve unreliable diagnostic classification on the large-scale aggregated data sets. In this work, we propose a novel graph-based classification model using the deep belief network (DBN) and the Autism Brain Imaging Data Exchange (ABIDE) database, which is a worldwide multisite functional and structural brain imaging data aggregation. The remarkable connectivity features are selected through a graph extension of K -nearest neighbors and then refined by a restricted path-based depth-first search algorithm. Thanks to the feature reduction, lower computational complexity could contribute to the shortening of the training time. The automatic hyperparameter-tuning technique is introduced to optimize the hyperparameters of the DBN by exploring the potential parameter space. The simulation experiments demonstrate the superior performance of our model, which is 6.4% higher than the best result reported on the ABIDE database. We also propose to use the data augmentation and the oversampling technique to identify further the possible subtypes within the ASD. The interpretability of our model enables the identification of the most remarkable autistic neural correlation patterns from the data-driven outcomes.

Brain function and clinical characterization in the Boston adolescent neuroimaging of depression and anxiety study.

We present a Human Connectome Project study tailored toward adolescent anxiety and depression. This study is one of the first studies of the Connectomes Related to Human Diseases initiative and is collecting structural, functional, and diffusion-weighted brain imaging data from up to 225 adolescents (ages 14-17 years), 150 of whom are expected to have a current diagnosis of an anxiety and/or depressive disorder. Comprehensive clinical and neuropsychological evaluations and longitudinal clinical data are also being collected. This article provides an overview of task functional magnetic resonance imaging (fMRI) protocols and preliminary findings (N=140), as well as clinical and neuropsychological characterization of adolescents. Data collection is ongoing for an additional 85 adolescents, most of whom are expected to have a diagnosis of an anxiety and/or depressive disorder. Data from the first 140 adolescents are projected for public release through the National Institutes of Health Data Archive (NDA) with the timing of this manuscript. All other data will be made publicly-available through the NDA at regularly scheduled intervals. This article is intended to serve as an introduction to this project as well as a reference for those seeking to clinical, neurocognitive, and task fMRI data from this public resource.

Decoding Task-Related Functional Brain Imaging Data to Identify Developmental Disorders: The Case of Congenital Amusia.

Machine learning classification techniques are frequently applied to structural and resting-state fMRI data to identify brain-based biomarkers for developmental disorders. However, task-related fMRI has rarely been used as a diagnostic tool. Here, we used structural MRI, resting-state connectivity and task-based fMRI data to detect congenital amusia, a pitch-specific developmental disorder. All approaches discriminated amusics from controls in meaningful brain networks at similar levels of accuracy. Interestingly, the classifier outcome was specific to deficit-related neural circuits, as the group classification failed for fMRI data acquired during a verbal task for which amusics were unimpaired. Most importantly, classifier outputs of task-related fMRI data predicted individual behavioral performance on an independent pitch-based task, while this relationship was not observed for structural or resting-state data. These results suggest that task-related imaging data can potentially be used as a powerful diagnostic tool to identify developmental disorders as they allow for the prediction of symptom severity.