Deep neural networks and kernel regression achieve comparable accuracies for functional connectivity prediction of behavior and demographics

A multicohort geometric deep learning study of age dependent cortical and subcortical morphologic interactions for fluid intelligence prediction

The relationship of human brain structure to cognitive function is complex, and how this relationship differs between childhood and adulthood is poorly understood. One strong hypothesis suggests the cognitive function of Fluid Intelligence (Gf) is dependent on prefrontal cortex and parietal cortex. In this work, we developed a novel graph convolutional neural networks (gCNNs) for the analysis of localized anatomic shape and prediction of Gf. Morphologic information of the cortical ribbons and subcortical structures was extracted from T1-weighted MRIs within two independent cohorts, the Adolescent Brain Cognitive Development Study (ABCD; age: 9.93 ± 0.62 years) of children and the Human Connectome Project (HCP; age: 28.81 ± 3.70 years). Prediction combining cortical and subcortical surfaces together yielded the highest accuracy of Gf for both ABCD (R = 0.314) and HCP datasets (R = 0.454), outperforming the state-of-the-art prediction of Gf from any other brain measures in the literature. Across both datasets, the morphology of the amygdala, hippocampus, and nucleus accumbens, along with temporal, parietal and cingulate cortex consistently drove the prediction of Gf, suggesting a significant reframing of the relationship between brain morphology and Gf to include systems involved with reward/aversion processing, judgment and decision-making, motivation, and emotion.

Proportional intracranial volume correction differentially biases behavioral predictions across neuroanatomical features, sexes, and development

Individual differences in brain anatomy can be used to predict variations in cognitive ability. Most studies to date have focused on broad population-level trends, but the extent to which the observed predictive features are shared across sexes and age groups remains to be established. While it is standard practice to account for intracranial volume (ICV) using proportion correction in both regional and whole-brain morphometric analyses, in the context of brain-behavior predictions the possible differential impact of ICV correction on anatomical features and subgroups within the population has yet to be systematically investigated. In this work, we evaluate the effect of proportional ICV correction on sex-independent and sex-specific predictive models of individual cognitive abilities across multiple anatomical properties (surface area, gray matter volume, and cortical thickness) in healthy young adults (Human Connectome Project; n = 1013, 548 females) and typically developing children (Adolescent Brain Cognitive Development study; n = 1823, 979 females). We demonstrate that ICV correction generally reduces predictive accuracies derived from surface area and gray matter volume, while increasing predictive accuracies based on cortical thickness in both adults and children. Furthermore, the extent to which predictive models generalize across sexes and age groups depends on ICV correction: models based on surface area and gray matter volume are more generalizable without ICV correction, while models based on cortical thickness are more generalizable with ICV correction. Finally, the observed neuroanatomical features predictive of cognitive abilities are unique across age groups regardless of ICV correction, but whether they are shared or unique across sexes (within age groups) depends on ICV correction. These findings highlight the importance of considering individual differences in ICV, and show that proportional ICV correction does not remove the effects of cranial volume from anatomical measurements and can introduce ICV bias where previously there was none. ICV correction choices affect not just the strength of the relationships captured, but also the conclusions drawn regarding the neuroanatomical features that underlie those relationships.

Comparison of individualized behavioral predictions across anatomical, diffusion and functional connectivity MRI

A fundamental goal across the neurosciences is the characterization of relationships linking brain anatomy, functioning, and behavior. Although various MRI modalities have been developed to probe these relationships, direct comparisons of their ability to predict behavior have been lacking. Here, we compared the ability of anatomical T1, diffusion and functional MRI (fMRI) to predict behavior at an individual level. Cortical thickness, area and volume were extracted from anatomical T1 images. Diffusion Tensor Imaging (DTI) and approximate Neurite Orientation Dispersion and Density Imaging (NODDI) models were fitted to the diffusion images. The resulting metrics were projected to the Tract-Based Spatial Statistics (TBSS) skeleton. We also ran probabilistic tractography for the diffusion images, from which we extracted the stream count, average stream length, and the average of each DTI and NODDI metric across tracts connecting each pair of brain regions. Functional connectivity (FC) was extracted from both task and resting-state fMRI. Individualized prediction of a wide range of behavioral measures were performed using kernel ridge regression, linear ridge regression and elastic net regression. Consistency of the results were investigated with the Human Connectome Project (HCP) and Adolescent Brain Cognitive Development (ABCD) datasets. In both datasets, FC-based models gave the best prediction performance, regardless of regression model or behavioral measure. This was especially true for the cognitive component. Furthermore, all modalities were able to predict cognition better than other behavioral components. Combining all modalities improved prediction of cognition, but not other behavioral components. Finally, across all behaviors, combining resting and task FC yielded prediction performance similar to combining all modalities. Overall, our study suggests that in the case of healthy children and young adults, behaviorally-relevant information in T1 and diffusion features might reflect a subset of the variance captured by FC.

Statistical quantification of confounding bias in machine learning models

BACKGROUND

The lack of nonparametric statistical tests for confounding bias significantly hampers the development of robust, valid, and generalizable predictive models in many fields of research. Here I propose the partial confounder test, which, for a given confounder variable, probes the null hypotheses of the model being unconfounded.

RESULTS

The test provides a strict control for type I errors and high statistical power, even for nonnormally and nonlinearly dependent predictions, often seen in machine learning. Applying the proposed test on models trained on large-scale functional brain connectivity data (N= 1,865) (i) reveals previously unreported confounders and (ii) shows that state-of-the-art confound mitigation approaches may fail preventing confounder bias in several cases.

CONCLUSIONS

The proposed test (implemented in the package mlconfound; https://mlconfound.readthedocs.io) can aid the assessment and improvement of the generalizability and validity of predictive models and, thereby, fosters the development of clinically useful machine learning biomarkers.

Decoding working memory-related information from repeated psychophysiological EEG experiments using convolutional and contrastive neural networks

OBJECTIVE

Extracting reliable information from EEG signals is difficult because the low signal-to-noise ratio and significant intersubject variability seriously hinder statistical analyses. However, recent advances in explainable machine learning open a new strategy to address this problem.

APPROACH

The current study evaluates this approach using results from the classification and decoding of electrical brain activity associated with information retention. We designed four neural network models differing in architecture, training strategies, and input representation to classify single experimental trials of a working memory task.

MAIN RESULTS

Our best models achieved an accuracy of 65.29$±0.76 and Matthews correlation coefficient of 0.288±0.018, outperforming the reference model trained on the same data. The highest correlation between classification score and behavioral performance was 0.36 (p=0.0007). Using analysis of input perturbation, we estimated the importance of EEG channels and frequency bands in the task at hand. The set of essential features identified for each network varies. We identified a subset of features common to all models that identified brain regions and frequency bands consistent with current neurophysiological knowledge of the processes critical to attention and working memory. Finally, we proposed sanity checks to examine further the robustness of each model's set of features.

SIGNIFICANCE

Our results indicate that explainable deep learning is a powerful tool for decoding information from EEG signals. It is crucial to train and analyze a range of models to identify stable and reliable features. Our results highlight the need for explainable modeling as the model with the highest accuracy appeared to use residual artifactual activity.

A cognitive state transformation model for task-general and task-specific subsystems of the brain connectome

The human brain flexibly controls different cognitive behaviors, such as memory and attention, to satisfy contextual demands. Much progress has been made to reveal task-induced modulations in the whole-brain functional connectome, but we still lack a way to model context-dependent changes. Here, we present a novel connectome-to-connectome (C2C) transformation framework that enables us to model the brain's functional reorganization from one connectome state to another in response to specific task goals. Using functional magnetic resonance imaging data from the Human Connectome Project, we demonstrate that the C2C model accurately generates an individual's task-related connectomes from their task-free (resting-state) connectome with a high degree of specificity across seven different cognitive states. Moreover, the C2C model amplifies behaviorally relevant individual differences in the task-free connectome, thereby improving behavioral predictions with increased power, achieving similar performance with just a third of the subjects needed when relying on resting-state data alone. Finally, the C2C model reveals how the brain reorganizes between cognitive states. Our observations support the existence of reliable state-specific subsystems in the brain and demonstrate that we can quantitatively model how the connectome reconfigures to different cognitive states, enabling more accurate predictions of behavior with fewer subjects.

Convolutional Recurrent Neural Network for Dynamic Functional MRI Analysis and Brain Disease Identification

Dynamic functional connectivity (dFC) networks derived from resting-state functional magnetic resonance imaging (rs-fMRI) help us understand fundamental dynamic characteristics of human brains, thereby providing an efficient solution for automated identification of brain diseases, such as Alzheimer's disease (AD) and its prodromal stage. Existing studies have applied deep learning methods to dFC network analysis and achieved good performance compared with traditional machine learning methods. However, they seldom take advantage of sequential information conveyed in dFC networks that could be informative to improve the diagnosis performance. In this paper, we propose a convolutional recurrent neural network (CRNN) for automated brain disease classification with rs-fMRI data. Specifically, we first construct dFC networks from rs-fMRI data using a sliding window strategy. Then, we employ three convolutional layers and long short-term memory (LSTM) layer to extract high-level features of dFC networks and also preserve the sequential information of extracted features, followed by three fully connected layers for brain disease classification. Experimental results on 174 subjects with 563 rs-fMRI scans from the Alzheimer's Disease Neuroimaging Initiative (ADNI) demonstrate the effectiveness of our proposed method in binary and multi-category classification tasks.

Robust Brain Age Estimation based on sMRI via Nonlinear Age-Adaptive Ensemble Learning

Precise prediction on brain age is urgently needed by many biomedical areas including mental rehabilitation prognosis as well as various medicine or treatment trials. People began to realize that contrasting physical (real) age and predicted brain age can help to highlight brain issues and evaluate if patients' brains are healthy or not. Such age prediction is often challenging for single model-based prediction, while the conditions of brains vary drastically over age. In this work, we present an age-adaptive ensemble model that is based on the combination of four different machine learning algorithms, including a support vector machine (SVR), a convolutional neural network (CNN) model, and the popular GoogLeNet and ResNet deep networks. The ensemble model proposed here is nonlinearly adaptive, where age is taken as a key factor in the nonlinear combination of various single-algorithm-based independent models. In our age-adaptive ensemble method, the weights of each model are learned automatically as nonlinear functions over age instead of fixed values, while brain age estimation is based on such an age-adaptive integration of various single models. The quality of the model is quantified by the mean absolute errors (MAE) and spearman correlation between the predicted age and the actual age, with the least MAE and the highest Spearman correlation representing the highest accuracy in age prediction. By testing on the Predictive Analysis Challenge 2019 (PAC 2019) dataset, our novel ensemble model has achieved a MAE down to 3.19, which is a significantly increased accuracy in this brain age competition. If deployed in the real world, our novel ensemble model having an improved accuracy could potentially help doctors to identify the risk of brain diseases more accurately and quickly, thus helping pharmaceutical companies develop drugs or treatments precisely, and potential offer a new powerful tool for researchers in the field of brain science.

Building Models of Functional Interactions Among Brain Domains that Encode Varying Information Complexity: A Schizophrenia Case Study

Revealing associations among various structural and functional patterns of the brain can yield highly informative results about the healthy and disordered brain. Studies using neuroimaging data have more recently begun to utilize the information within as well as across various functional and anatomical domains (i.e., groups of brain networks). However, most whole-brain approaches assume similar complexity of interactions throughout the brain. Here we investigate the hypothesis that interactions between brain networks capture varying amounts of complexity, and that we can better capture this information by varying the complexity of the model subspace structure based on available training data. To do this, we employ a Bayesian optimization-based framework known as the Tree Parzen Estimator (TPE) to identify, exploit and analyze patterns of variation in the information encoded by temporal information extracted from functional magnetic resonance imaging (fMRI) subdomains of the brain. Using a repeated cross-validation procedure on a schizophrenia classification task, we demonstrate evidence that interactions between specific functional subdomains are better characterized by more sophisticated model architectures compared to less complicated ones required by the others for optimally contributing towards classification and understanding the brain's functional interactions. We show that functional subdomains known to be involved in schizophrenia require more complex architectures to optimally unravel discriminatory information about the disorder. Our study points to the need for adaptive, hierarchical learning frameworks that cater differently to the features from different subdomains, not only for a better prediction but also for enabling the identification of features predicting the outcome of interest.

Behavioral Studies Using Large-Scale Brain Networks – Methods and Validations

Mapping human behaviors to brain activity has become a key focus in modern cognitive neuroscience. As methods such as functional MRI (fMRI) advance cognitive scientists show an increasing interest in investigating neural activity in terms of functional connectivity and brain networks, rather than activation in a single brain region. Due to the noisy nature of neural activity, determining how behaviors are associated with specific neural signals is not well-established. Previous research has suggested graph theory techniques as a solution. Graph theory provides an opportunity to interpret human behaviors in terms of the topological organization of brain network architecture. Graph theory-based approaches, however, only scratch the surface of what neural connections relate to human behavior. Recently, the development of data-driven methods, e.g., machine learning and deep learning approaches, provide a new perspective to study the relationship between brain networks and human behaviors across the whole brain, expanding upon past literatures. In this review, we sought to revisit these data-driven approaches to facilitate our understanding of neural mechanisms and build models of human behaviors. We start with the popular graph theory approach and then discuss other data-driven approaches such as connectome-based predictive modeling, multivariate pattern analysis, network dynamic modeling, and deep learning techniques that quantify meaningful networks and connectivity related to cognition and behaviors. Importantly, for each topic, we discuss the pros and cons of the methods in addition to providing examples using our own data for each technique to describe how these methods can be applied to real-world neuroimaging data.

Meta-matching as a simple framework to translate phenotypic predictive models from big to small data

We propose a simple framework-meta-matching-to translate predictive models from large-scale datasets to new unseen non-brain-imaging phenotypes in small-scale studies. The key consideration is that a unique phenotype from a boutique study likely correlates with (but is not the same as) related phenotypes in some large-scale dataset. Meta-matching exploits these correlations to boost prediction in the boutique study. We apply meta-matching to predict non-brain-imaging phenotypes from resting-state functional connectivity. Using the UK Biobank (N = 36,848) and Human Connectome Project (HCP) (N = 1,019) datasets, we demonstrate that meta-matching can greatly boost the prediction of new phenotypes in small independent datasets in many scenarios. For example, translating a UK Biobank model to 100 HCP participants yields an eight-fold improvement in variance explained with an average absolute gain of 4.0% (minimum = -0.2%, maximum = 16.0%) across 35 phenotypes. With a growing number of large-scale datasets collecting increasingly diverse phenotypes, our results represent a lower bound on the potential of meta-matching.

Pattern learning reveals brain asymmetry to be linked to socioeconomic status

Socioeconomic status (SES) anchors individuals in their social network layers. Our embedding in the societal fabric resonates with habitus, world view, opportunity, and health disparity. It remains obscure how distinct facets of SES are reflected in the architecture of the central nervous system. Here, we capitalized on multivariate multi-output learning algorithms to explore possible imprints of SES in gray and white matter structure in the wider population (n ≈ 10,000 UK Biobank participants). Individuals with higher SES, compared with those with lower SES, showed a pattern of increased region volumes in the left brain and decreased region volumes in the right brain. The analogous lateralization pattern emerged for the fiber structure of anatomical white matter tracts. Our multimodal findings suggest hemispheric asymmetry as an SES-related brain signature, which was consistent across six different indicators of SES: degree, education, income, job, neighborhood and vehicle count. Hence, hemispheric specialization may have evolved in human primates in a way that reveals crucial links to SES.

Reporting details of neuroimaging studies on individual traits prediction: A literature survey

Using machine-learning tools to predict individual phenotypes from neuroimaging data is one of the most promising and hence dynamic fields in systems neuroscience. Here, we perform a literature survey of the rapidly work on phenotype prediction in healthy subjects or general population to sketch out the current state and ongoing developments in terms of data, analysis methods and reporting. Excluding papers on age-prediction and clinical applications, which form a distinct literature, we identified a total 108 papers published since 2007. In these, memory, fluid intelligence and attention were most common phenotypes to be predicted, which resonates with the observation that roughly a quarter of the papers used data from the Human Connectome Project, even though another half recruited their own cohort. Sample size (in terms of training and external test sets) and prediction accuracy (from internal and external validation respectively) did not show significant temporal trends. Prediction accuracy was negatively correlated with sample size of the training set, but not the external test set. While known to be optimistic, leave-one-out cross-validation (LOO CV) was the prevalent strategy for model validation (n = 48). Meanwhile, 27 studies used external validation with external test set. Both numbers showed no significant temporal trends. The most popular learning algorithm was connectome-based predictive modeling introduced by the Yale team. Other common learning algorithms were linear regression, relevance vector regression (RVR), support vector regression (SVR), least absolute shrinkage and selection operator (LASSO), and elastic net. Meanwhile, the amount of data from self-recruiting studies (but not studies using open, shared dataset) was positively correlated with internal validation prediction accuracy. At the same time, self-recruiting studies also reported a significantly higher internal validation prediction accuracy than those using open, shared datasets. Data type and participant age did not significantly influence prediction accuracy. Confound control also did not influence prediction accuracy after adjusted for other factors. To conclude, most of the current literature is probably quite optimistic with internal validation using LOO CV. More efforts should be made to encourage the use of external validation with external test sets to further improve generalizability of the models.

Machine Learning Evidence for Sex Differences Consistently Influences Resting-State Functional Magnetic Resonance Imaging Fluctuations Across Multiple Independently Acquired Data Sets

Background/Introduction: Sex classification using functional connectivity from resting-state functional magnetic resonance imaging (rs-fMRI) has shown promising results. This suggested that sex difference might also be embedded in the blood-oxygen-level-dependent properties such as the amplitude of low-frequency fluctuation (ALFF) and the fraction of ALFF (fALFF). This study comprehensively investigates sex differences using a reliable and explainable machine learning (ML) pipeline. Five independent cohorts of rs-fMRI with over than 5500 samples were used to assess sex classification performance and map the spatial distribution of the important brain regions. Methods: Five rs-fMRI samples were used to extract ALFF and fALFF features from predefined brain parcellations and then were fed into an unbiased and explainable ML pipeline with a wide range of methods. The pipeline comprehensively assessed unbiased performance for within-sample and across-sample validation. In addition, the parcellation effect, classifier selection, scanning length, spatial distribution, reproducibility, and feature importance were analyzed and evaluated thoroughly in the study. Results: The results demonstrated high sex classification accuracies from healthy adults (area under the curve >0.89), while degrading for nonhealthy subjects. Sex classification showed moderate to good intraclass correlation coefficient based on parcellation. Linear classifiers outperform nonlinear classifiers. Sex differences could be detected even with a short rs-fMRI scan (e.g., 2 min). The spatial distribution of important features overlaps with previous results from studies. Discussion: Sex differences are consistent in rs-fMRI and should be considered seriously in any study design, analysis, or interpretation. Features that discriminate males and females were found to be distributed across several different brain regions, suggesting a complex mosaic for sex differences in rs-fMRI. Impact statement The presented study unraveled that sex differences are embedded in the blood-oxygen-level dependent (BOLD) and can be predicted using unbiased and explainable machine learning pipeline. The study revealed that psychiatric disorders and demographics might influence the BOLD signal and interact with the classification of sex. The spatial distribution of the important features presented here supports the notion that the brain is a mosaic of male and female features. The findings emphasize the importance of controlling for sex when conducting brain imaging analysis. In addition, the presented framework can be adapted to classify other variables from resting-state BOLD signals.

Methodological evaluation of individual cognitive prediction based on the brain white matter structural connectome

An emerging trend is to use regression‐based machine learning approaches to predict cognitive functions at the individual level from neuroimaging data. However, individual prediction models are inherently influenced by the vast options for network construction and model selection in machine learning pipelines. In particular, the brain white matter (WM) structural connectome lacks a systematic evaluation of the effects of different options in the pipeline on predictive performance. Here, we focused on the methodological evaluation of brain structural connectome‐based predictions. For network construction, we considered two parcellation schemes for defining nodes and seven strategies for defining edges. For the regression algorithms, we used eight regression models. Four cognitive domains and brain age were targeted as predictive tasks based on two independent datasets (Beijing Aging Brain Rejuvenation Initiative [BABRI]: 633 healthy older adults; Human Connectome Projects in Aging [HCP‐A]: 560 healthy older adults). Based on the results, the WM structural connectome provided a satisfying predictive ability for individual age and cognitive functions, especially for executive function and attention. Second, different parcellation schemes induce a significant difference in predictive performance. Third, prediction results from different data sets showed that dMRI with distinct acquisition parameters may plausibly result in a preference for proper fiber reconstruction algorithms and different weighting options. Finally, deep learning and Elastic‐Net models are more accurate and robust in connectome‐based predictions. Together, significant effects of different options in WM network construction and regression algorithms on the predictive performances are identified in this study, which may provide important references and guidelines to select suitable options for future studies in this field.

Shared and unique brain network features predict cognitive, personality, and mental health scores in the ABCD study

How individual differences in brain network organization track behavioral variability is a fundamental question in systems neuroscience. Recent work suggests that resting-state and task-state functional connectivity can predict specific traits at the individual level. However, most studies focus on single behavioral traits, thus not capturing broader relationships across behaviors. In a large sample of 1858 typically developing children from the Adolescent Brain Cognitive Development (ABCD) study, we show that predictive network features are distinct across the domains of cognitive performance, personality scores and mental health assessments. On the other hand, traits within each behavioral domain are predicted by similar network features. Predictive network features and models generalize to other behavioral measures within the same behavioral domain. Although tasks are known to modulate the functional connectome, predictive network features are similar between resting and task states. Overall, our findings reveal shared brain network features that account for individual variation within broad domains of behavior in childhood.

The trait and state negative affect can be separately predicted by stable and variable resting-state functional connectivity

BACKGROUND

Many emotional experiences such as anxiety and depression are influenced by negative affect (NA). NA has both trait and state features, which play different roles in physiological and mental health. Attending to NA common to various emotional experiences and their trait-state features might help deepen the understanding of the shared foundation of related emotional disorders.

METHODS

The principal component of five measures was calculated to indicate individuals' NA level. Applying the connectivity-based correlation analysis, we first identified resting-state functional connectives (FCs) relating to NA in sample 1 (n = 367), which were validated through an independent sample (n = 232; sample 2). Next, based on the variability of FCs across large timescale, we further divided the NA-related FCs into high- and low-variability groups. Finally, FCs in different variability groups were separately applied to predict individuals' neuroticism level (which is assumed to be the core trait-related factor underlying NA), and the change of NA level (which represents the state-related fluctuation of NA).

RESULTS

The low-variability FCs were primarily within the default mode network (DMN) and between the DMN and dorsal attention network/sensory system and significantly predicted trait rather than state NA. The high-variability FCs were primarily between the DMN and ventral attention network, the fronto-parietal network and DMN/sensory system, and significantly predicted the change of NA level.

CONCLUSIONS

The trait and state NA can be separately predicted by stable and variable spontaneous FCs with different attentional processes and emotion regulatory mechanisms, which could deepen our understanding of NA.

Cross-ethnicity/race generalization failure of behavioral prediction from resting-state functional connectivity

Algorithmic biases that favor majority populations pose a key challenge to the application of machine learning for precision medicine. Here, we assessed such bias in prediction models of behavioral phenotypes from brain functional magnetic resonance imaging. We examined the prediction bias using two independent datasets (preadolescent versus adult) of mixed ethnic/racial composition. When predictive models were trained on data dominated by white Americans (WA), out-of-sample prediction errors were generally higher for African Americans (AA) than for WA. This bias toward WA corresponds to more WA-like brain-behavior association patterns learned by the models. When models were trained on AA only, compared to training only on WA or an equal number of AA and WA participants, AA prediction accuracy improved but stayed below that for WA. Overall, the results point to the need for caution and further research regarding the application of current brain-behavior prediction models in minority populations.

ConCeptCNN: A novel multi-filter convolutional neural network for the prediction of neurodevelopmental disorders using brain connectome

BACKGROUND

Deep convolutional neural network (CNN) and its derivatives have recently shown great promise in the prediction of brain disorders using brain connectome data. Existing deep CNN methods using single global row and column convolutional filters have limited ability to extract discriminative information from brain connectome for prediction tasks.

PURPOSE

This paper presents a novel deep Connectome-Inception CNN (ConCeptCNN) model, which is developed based on multiple convolutional filters. The proposed model is used to extract topological features from brain connectome data for neurological disorders classification and analysis.

METHODS

The ConCeptCNN uses multiple vector-shaped filters extract topological information from the brain connectome at different levels for complementary feature embeddings of brain connectome. The proposed model is validated using two datasets: the Neuro Bureau ADHD-200 dataset and the Cincinnati Early Prediction Study (CINEPS) dataset.

RESULTS

In a cross-validation experiment, the ConCeptCNN achieved a prediction accuracy of 78.7% for the detection of ADHD in adolescents and an accuracy of 81.6% for the prediction of cognitive deficits at 2-year corrected age in very preterm infants. In addition to the classification tasks, the ConCeptCNN identified several brain regions that are discriminative to neurodevelopmental disorders.

CONCLUSIONS

We compared the ConCeptCNN with several peer CNN methods. The results demonstrated that proposed model improves overall classification performance of neurodevelopmental disorders prediction tasks. This article is protected by copyright. All rights reserved.

Performance scaling for structural MRI surface parcellations: a machine learning analysis in the ABCD Study

The use of predefined parcellations on surface-based representations of the brain as a method for data reduction is common across neuroimaging studies. In particular, prediction-based studies typically employ parcellation-driven summaries of brain measures as input to predictive algorithms, but the choice of parcellation and its influence on performance is often ignored. Here we employed preprocessed structural magnetic resonance imaging (sMRI) data from the Adolescent Brain Cognitive Development Study® to examine the relationship between 220 parcellations and out-of-sample predictive performance across 45 phenotypic measures in a large sample of 9- to 10-year-old children (N = 9,432). Choice of machine learning (ML) pipeline and use of alternative multiple parcellation-based strategies were also assessed. Relative parcellation performance was dependent on the spatial resolution of the parcellation, with larger number of parcels (up to ~4,000) outperforming coarser parcellations, according to a power-law scaling of between 1/4 and 1/3. Performance was further influenced by the type of parcellation, ML pipeline, and general strategy, with existing literature-based parcellations, a support vector-based pipeline, and ensembling across multiple parcellations, respectively, as the highest performing. These findings highlight the choice of parcellation as an important influence on downstream predictive performance, showing in some cases that switching to a higher resolution parcellation can yield a relatively large boost to performance.

Decoding task specific and task general functional architectures of the brain

Functional magnetic resonance imaging (fMRI) is used to capture complex and dynamic interactions between brain regions while performing tasks. Task related alterations in the brain have been classified as task specific and task general, depending on whether they are particular to a task or common across multiple tasks. Using recent attempts in interpreting deep learning models, we propose an approach to determine both task specific and task general architectures of the functional brain. We demonstrate our methods with a reference‐based decoder on deep learning classifiers trained on 12,500 rest and task fMRI samples from the Human Connectome Project (HCP). The decoded task general and task specific motor and language architectures were validated with findings from previous studies. We found that unlike intersubject variability that is characteristic of functional pathology of neurological diseases, a small set of connections are sufficient to delineate the rest and task states. The nodes and connections in the task general architecture could serve as potential disease biomarkers as alterations in task general brain modulations are known to be implicated in several neuropsychiatric disorders.

Attention module improves both performance and interpretability of four-dimensional functional magnetic resonance imaging decoding neural network

Decoding brain cognitive states from neuroimaging signals is an important topic in neuroscience. In recent years, deep neural networks (DNNs) have been recruited for multiple brain state decoding and achieved good performance. However, the open question of how to interpret the DNN black box remains unanswered. Capitalizing on advances in machine learning, we integrated attention modules into brain decoders to facilitate an in‐depth interpretation of DNN channels. A four‐dimensional (4D) convolution operation was also included to extract temporo‐spatial interaction within the fMRI signal. The experiments showed that the proposed model obtains a very high accuracy (97.4%) and outperforms previous researches on the seven different task benchmarks from the Human Connectome Project (HCP) dataset. The visualization analysis further illustrated the hierarchical emergence of task‐specific masks with depth. Finally, the model was retrained to regress individual traits within the HCP and to classify viewing images from the BOLD5000 dataset, respectively. Transfer learning also achieves good performance. Further visualization analysis shows that, after transfer learning, low‐level attention masks remained similar to the source domain, whereas high‐level attention masks changed adaptively. In conclusion, the proposed 4D model with attention module performed well and facilitated interpretation of DNNs, which is helpful for subsequent research.

Shared functional connections within and between cortical networks predict cognitive abilities in adult males and females

A thorough understanding of sex-independent and sex-specific neurobiological features that underlie cognitive abilities in healthy individuals is essential for the study of neurological illnesses in which males and females differentially experience and exhibit cognitive impairment. Here, we evaluate sex-independent and sex-specific relationships between functional connectivity and individual cognitive abilities in 392 healthy young adults (196 males) from the Human Connectome Project. First, we establish that sex-independent models comparably predict crystallised abilities in males and females, but only successfully predict fluid abilities in males. Second, we demonstrate sex-specific models comparably predict crystallised abilities within and between sexes, and generally fail to predict fluid abilities in either sex. Third, we reveal that largely overlapping connections between visual, dorsal attention, ventral attention, and temporal parietal networks are associated with better performance on crystallised and fluid cognitive tests in males and females, while connections within visual, somatomotor, and temporal parietal networks are associated with poorer performance. Together, our findings suggest that shared neurobiological features of the functional connectome underlie crystallised and fluid abilities across the sexes.

Predicting Individual Task Contrasts From Resting-state Functional Connectivity using a Surface-based Convolutional Network

Task-based and resting-state represent the two most common experimental paradigms of functional neuroimaging. While resting-state offers a flexible and scalable approach for characterizing brain function, task-based techniques provide superior localization. In this paper, we build on recent deep learning methods to create a model that predicts task-based contrast maps from resting-state fMRI scans. Specifically, we propose BrainSurfCNN, a surface-based fully-convolutional neural network model that works with a representation of the brain's cortical sheet. BrainSurfCNN achieves exceptional predictive accuracy on independent test data from the Human Connectome Project, which is on par with the repeat reliability of the measured subject-level contrast maps. Conversely, our analyses reveal that a previously published benchmark is no better than group-average contrast maps. Finally, we demonstrate that BrainSurfCNN can generalize remarkably well to novel domains with limited training data.

A deep learning based approach identifies regions more relevant than resting-state networks to the prediction of general intelligence from resting-state fMRI

Prediction of cognitive ability latent factors such as general intelligence from neuroimaging has elucidated questions pertaining to their neural origins. However, predicting general intelligence from functional connectivity limit hypotheses to that specific domain, being agnostic to time‐distributed features and dynamics. We used an ensemble of recurrent neural networks to circumvent this limitation, bypassing feature extraction, to predict general intelligence from resting‐state functional magnetic resonance imaging regional signals of a large sample (n = 873) of Human Connectome Project adult subjects. Ablating common resting‐state networks (RSNs) and measuring degradation in performance, we show that model reliance can be mostly explained by network size. Using our approach based on the temporal variance of saliencies, that is, gradients of outputs with regards to inputs, we identify a candidate set of networks that more reliably affect performance in the prediction of general intelligence than similarly sized RSNs. Our approach allows us to further test the effect of local alterations on data and the expected changes in derived metrics such as functional connectivity and instantaneous innovations.

Predicting behavior through dynamic modes in resting-state fMRI data

Dynamic properties of resting-state functional connectivity (FC) provide rich information on brain-behavior relationships. Dynamic mode decomposition (DMD) has been used as a method to characterize FC dynamics. However, it remains unclear whether dynamic modes (DMs), spatial-temporal coherent patterns computed by DMD, provide information about individual behavioral differences. This study established a methodological approach to predict individual differences in behavior using DMs. Furthermore, we investigated the contribution of DMs within each of seven specific frequency bands (0-0.1,...,0.6-0.7 Hz) for prediction. To validate our approach, we tested whether each of 59 behavioral measures could be predicted by performing multivariate pattern analysis on a Gram matrix, which was created using subject-specific DMs computed from resting-state functional magnetic resonance imaging (rs-fMRI) data of individuals. DMD successfully predicted behavior and outperformed temporal and spatial independent component analysis, which is the conventional data decomposition method for extracting spatial activity patterns. Most of the behavioral measures that were predicted with significant accuracy in a permutation test were related to cognition. We found that DMs within frequency bands <0.2 Hz primarily contributed to prediction and had spatial structures similar to several common resting-state networks. Our results indicate that DMD is efficient in extracting spatiotemporal features from rs-fMRI data.

Predicting cognitive scores with graph neural networks through sample selection learning

Analyzing the relation between intelligence and neural activity is of the utmost importance in understanding the working principles of the human brain in health and disease. In existing literature, functional brain connectomes have been used successfully to predict cognitive measures such as intelligence quotient (IQ) scores in both healthy and disordered cohorts using machine learning models. However, existing methods resort to flattening the brain connectome (i.e., graph) through vectorization which overlooks its topological properties. To address this limitation and inspired from the emerging graph neural networks (GNNs), we design a novel regression GNN model (namely RegGNN) for predicting IQ scores from brain connectivity. On top of that, we introduce a novel, fully modular sample selection method to select the best samples to learn from for our target prediction task. However, since such deep learning architectures are computationally expensive to train, we further propose a learning-based sample selection method that learns how to choose the training samples with the highest expected predictive power on unseen samples. For this, we capitalize on the fact that connectomes (i.e., their adjacency matrices) lie in the symmetric positive definite (SPD) matrix cone. Our results on full-scale and verbal IQ prediction outperforms comparison methods in autism spectrum disorder cohorts and achieves a competitive performance for neurotypical subjects using 3-fold cross-validation. Furthermore, we show that our sample selection approach generalizes to other learning-based methods, which shows its usefulness beyond our GNN architecture.

Deep Neural Networks to Detect Weeds from Crops in Agricultural Environments in Real-Time: A Review

Automation, including machine learning technologies, are becoming increasingly crucial in agriculture to increase productivity. Machine vision is one of the most popular parts of machine learning and has been widely used where advanced automation and control have been required. The trend has shifted from classical image processing and machine learning techniques to modern artificial intelligence (AI) and deep learning (DL) methods. Based on large training datasets and pre-trained models, DL-based methods have proven to be more accurate than previous traditional techniques. Machine vision has wide applications in agriculture, including the detection of weeds and pests in crops. Variation in lighting conditions, failures to transfer learning, and object occlusion constitute key challenges in this domain. Recently, DL has gained much attention due to its advantages in object detection, classification, and feature extraction. DL algorithms can automatically extract information from large amounts of data used to model complex problems and is, therefore, suitable for detecting and classifying weeds and crops. We present a systematic review of AI-based systems to detect weeds, emphasizing recent trends in DL. Various DL methods are discussed to clarify their overall potential, usefulness, and performance. This study indicates that several limitations obstruct the widespread adoption of AI/DL in commercial applications. Recommendations for overcoming these challenges are summarized.

Deep learning in resting-state fMRI. ANNUAL INTERNATIONAL CONFERENCE OF THE IEEE ENGINEERING IN MEDICINE AND BIOLOGY SOCIETY. IEEE ENGINEERING IN MEDICINE AND BIOLOGY SOCIETY. ANNUAL INTERNATIONAL CONFERENCE 2021;2021:3965-3969. [PMID: 34892099 DOI: 10.1109/embc46164.2021.9630257]

Modeling the rich, dynamic spatiotemporal variations captured by human brain functional magnetic resonance imaging (fMRI) data is a complicated task. Analysis at the brain's regional and connection levels provides more straightforward biological interpretation for fMRI data and has been instrumental in characterizing the brain thus far. Here we hypothesize that spatiotemporal learning directly in the four-dimensional (4D) fMRI voxel-time space could result in enhanced discriminative brain representations compared to widely used, pre-engineered fMRI temporal transformations, and brain regional and connection-level fMRI features. Motivated by this, we extend our recently reported structural MRI (sMRI) deep learning (DL) pipeline to additionally capture temporal variations, training the proposed 4D DL model end-to-end on preprocessed fMRI data. Results validate that the complex non-linear functions of the used deep spatiotemporal approach generate discriminative encodings for the studied learning task, outperforming both standard machine learning (SML) and DL methods on the widely used fMRI voxel/region/connection features, except the relatively simplistic measure of central tendency - the temporal mean of the fMRI data. Additionally, we identify the fMRI features for which DL significantly outperformed SML methods for voxel-level fMRI features. Overall, our results support the efficiency and potential of DL models trainable at the voxel level fMRI data and highlight the importance of developing auxiliary tools to facilitate interpretation of such flexible models.

Representation learning of resting state fMRI with variational autoencoder

Resting state functional magnetic resonance imaging (rsfMRI) data exhibits complex but structured patterns. However, the underlying origins are unclear and entangled in rsfMRI data. Here we establish a variational auto-encoder, as a generative model trainable with unsupervised learning, to disentangle the unknown sources of rsfMRI activity. After being trained with large data from the Human Connectome Project, the model has learned to represent and generate patterns of cortical activity and connectivity using latent variables. The latent representation and its trajectory represent the spatiotemporal characteristics of rsfMRI activity. The latent variables reflect the principal gradients of the latent trajectory and drive activity changes in cortical networks. Representational geometry captured as covariance or correlation between latent variables, rather than cortical connectivity, can be used as a more reliable feature to accurately identify subjects from a large group, even if only a short period of data is available in each subject. Our results demonstrate that VAE is a valuable addition to existing tools, particularly suited for unsupervised representation learning of resting state fMRI activity.

Computing schizophrenia: ethical challenges for machine learning in psychiatry

Recent advances in machine learning (ML) promise far-reaching improvements across medical care, not least within psychiatry. While to date no psychiatric application of ML constitutes standard clinical practice, it seems crucial to get ahead of these developments and address their ethical challenges early on. Following a short general introduction concerning ML in psychiatry, we do so by focusing on schizophrenia as a paradigmatic case. Based on recent research employing ML to further the diagnosis, treatment, and prediction of schizophrenia, we discuss three hypothetical case studies of ML applications with view to their ethical dimensions. Throughout this discussion, we follow the principlist framework by Tom Beauchamp and James Childress to analyse potential problems in detail. In particular, we structure our analysis around their principles of beneficence, non-maleficence, respect for autonomy, and justice. We conclude with a call for cautious optimism concerning the implementation of ML in psychiatry if close attention is paid to the particular intricacies of psychiatric disorders and its success evaluated based on tangible clinical benefit for patients.

Machine learning prediction of cognition from functional connectivity: Are feature weights reliable?

Cognitive performance can be predicted from an individual's functional brain connectivity with modest accuracy using machine learning approaches. As yet, however, predictive models have arguably yielded limited insight into the neurobiological processes supporting cognition. To do so, feature selection and feature weight estimation need to be reliable to ensure that important connections and circuits with high predictive utility can be reliably identified. We comprehensively investigate feature weight test-retest reliability for various predictive models of cognitive performance built from resting-state functional connectivity networks in healthy young adults (n=400). Despite achieving modest prediction accuracies (r=0.2-0.4), we find that feature weight reliability is generally poor for all predictive models (ICC< 0.3), and significantly poorer than predictive models for overt biological attributes such as sex (ICC≈0.5). Larger sample sizes (n=800), the Haufe transformation, non-sparse feature selection/regularization and smaller feature spaces marginally improve reliability (ICC< 0.4). We elucidate a tradeoff between feature weight reliability and prediction accuracy and find that univariate statistics are marginally more reliable than feature weights from predictive models. Finally, we show that measuring agreement in feature weights between cross-validation folds provides inflated estimates of feature weight reliability. We thus recommend for reliability to be estimated out-of-sample, if possible. We argue that rebalancing focus from prediction accuracy to model reliability may facilitate mechanistic understanding of cognition with machine learning approaches.

Advanced brain ageing in Parkinson's disease is related to disease duration and individual impairment

Machine learning can reliably predict individual age from MRI data, revealing that patients with neurodegenerative disorders show an elevated biological age. A surprising gap in the literature, however, pertains to Parkinson's disease. Here, we evaluate brain age in two cohorts of Parkinson's patients and investigated the relationship between individual brain age and clinical characteristics. We assessed 372 patients with idiopathic Parkinson's disease, newly diagnosed cases from the Parkinson's Progression Marker Initiative database and a more chronic local sample, as well as age- and sex-matched healthy controls. Following morphometric preprocessing and atlas-based compression, individual brain age was predicted using a multivariate machine learning model trained on an independent, multi-site reference sample. Across cohorts, healthy controls were well predicted with a mean error of 4.4 years. In turn, Parkinson's patients showed a significant (controlling for age, gender and site) increase in brain age of ∼3 years. While this effect was already present in the newly diagnosed sample, advanced biological age was significantly related to disease duration as well as worse cognitive and motor impairment. While biological age is increased in patients with Parkinson's disease, the effect is at the lower end of what is found for other neurological and psychiatric disorders. We argue that this may reflect a heterochronicity between forebrain atrophy and small but behaviourally salient midbrain pathology. Finally, we point to the need to disentangle physiological ageing trajectories, lifestyle effects and core pathological changes.

Connectome-based individual prediction of cognitive behaviors via graph propagation network reveals directed brain network topology

Objective. Brain connectivity network supports the information flow underlying human cognitions and should reflect the individual variability in human cognitive behaviors. Various studies have utilized brain connectivity to predict individual differences in human behaviors. However, traditional studies viewed brain connectivity network as a one-dimensional vector, a method which neglects topological properties of brain connectivity network.Approach. To utilize these topological properties, we proposed that graph neural network (GNN) which combines graph theory and neural network can be adopted. Different from previous node-driven GNNs that parameterize on the node feature transformation, we designed an edge-driven GNN named graph propagation network (GPN) that parameterizes on the information propagation within brain connectivity network.Main results.Edge-driven GPN outperforms various baseline models such as node-driven GNN and traditional partial least square regression in predicting the individual total cognition based on the resting-state functional connectome. GPN also reveals a directed network topology encoding the information flow, indicating that higher-order association cortices such as dorsolateral prefrontal, inferior frontal and inferior parietal cortices are responsible for the information integration underlying total cognition.Significance. These results suggest that edge-driven GPN can better explore topological structures of brain connectivity network and can serve as a new method to associate brain connectome and human behaviors.

Distinct functional and structural connections predict crystallised and fluid cognition in healthy adults

White matter pathways between neurons facilitate neuronal coactivation patterns in the brain. Insight into how these structural and functional connections underlie complex cognitive functions provides an important foundation with which to delineate disease‐related changes in cognitive functioning. Here, we integrate neuroimaging, connectomics, and machine learning approaches to explore how functional and structural brain connectivity relate to cognition. Specifically, we evaluate the extent to which functional and structural connectivity predict individual crystallised and fluid cognitive abilities in 415 unrelated healthy young adults (202 females) from the Human Connectome Project. We report three main findings. First, we demonstrate functional connectivity is more predictive of cognitive scores than structural connectivity, and, furthermore, integrating the two modalities does not increase explained variance. Second, we show the quality of cognitive prediction from connectome measures is influenced by the choice of grey matter parcellation, and, possibly, how that parcellation is derived. Third, we find that distinct functional and structural connections predict crystallised and fluid abilities. Taken together, our results suggest that functional and structural connectivity have unique relationships with crystallised and fluid cognition and, furthermore, studying both modalities provides a more comprehensive insight into the neural correlates of cognition.

Individual variation underlying brain age estimates in typical development

Typical brain development follows a protracted trajectory throughout childhood and adolescence. Deviations from typical growth trajectories have been implicated in neurodevelopmental and psychiatric disorders. Recently, the use of machine learning algorithms to model age as a function of structural or functional brain properties has been used to examine advanced or delayed brain maturation in healthy and clinical populations. Termed 'brain age', this approach often relies on complex, nonlinear models that can be difficult to interpret. In this study, we use model explanation methods to examine the cortical features that contribute to brain age modelling on an individual basis. In a large cohort of n = 768 typically-developing children (aged 3-21 years), we build models of brain development using three different machine learning approaches. We employ SHAP, a model-agnostic technique to identify sample-specific feature importance, to identify regional cortical metrics that explain errors in brain age prediction. We find that, on average, brain age prediction and the cortical features that explain model predictions are consistent across model types and reflect previously reported patterns of regions brain development. However, while several regions are found to contribute to brain age prediction error, we find little spatial correspondence between individual estimates of feature importance, even when matched for age, sex and brain age prediction error. We also find no association between brain age error and cognitive performance in this typically-developing sample. Overall, this study shows that, while brain age estimates based on cortical development are relatively robust and consistent across model types and preprocessing strategies, significant between-subject variation exists in the features that explain erroneous brain age predictions on an individual level.

Behavioural relevance of spontaneous, transient brain network interactions in fMRI

How spontaneously fluctuating functional magnetic resonance imaging (fMRI) signals in different brain regions relate to behaviour has been an open question for decades. Correlations in these signals, known as functional connectivity, can be averaged over several minutes of data to provide a stable representation of the functional network architecture for an individual. However, associations between these stable features and behavioural traits have been shown to be dominated by individual differences in anatomy. Here, using kernel learning tools, we propose methods to assess and compare the relation between time-varying functional connectivity, time-averaged functional connectivity, structural brain data, and non-imaging subject behavioural traits. We applied these methods to Human Connectome Project resting-state fMRI data to show that time-varying fMRI functional connectivity, detected at time-scales of a few seconds, has associations with some behavioural traits that are not dominated by anatomy. Despite time-averaged functional connectivity accounting for the largest proportion of variability in the fMRI signal between individuals, we found that some aspects of intelligence could only be explained by time-varying functional connectivity. The finding that time-varying fMRI functional connectivity has a unique relationship to population behavioural variability suggests that it might reflect transient neuronal communication fluctuating around a stable neural architecture.

Deep Neural Regression Prediction of Motor Imagery Skills Using EEG Functional Connectivity Indicators

Motor imaging (MI) induces recovery and neuroplasticity in neurophysical regulation. However, a non-negligible portion of users presents insufficient coordination skills of sensorimotor cortex control. Assessments of the relationship between wakefulness and tasks states are conducted to foster neurophysiological and mechanistic interpretation in MI-related applications. Thus, to understand the organization of information processing, measures of functional connectivity are used. Also, models of neural network regression prediction are becoming popular, These intend to reduce the need for extracting features manually. However, predicting MI practicing's neurophysiological inefficiency raises several problems, like enhancing network regression performance because of the overfitting risk. Here, to increase the prediction performance, we develop a deep network regression model that includes three procedures: leave-one-out cross-validation combined with Monte Carlo dropout layers, subject clustering of MI inefficiency, and transfer learning between neighboring runs. Validation is performed using functional connectivity predictors extracted from two electroencephalographic databases acquired in conditions close to real MI applications (150 users), resulting in a high prediction of pretraining desynchronization and initial training synchronization with adequate physiological interpretability.

Heritability of individualized cortical network topography

Human cortex is patterned by a complex and interdigitated web of large-scale functional networks. Recent methodological breakthroughs reveal variation in the size, shape, and spatial topography of cortical networks across individuals. While spatial network organization emerges across development, is stable over time, and is predictive of behavior, it is not yet clear to what extent genetic factors underlie interindividual differences in network topography. Here, leveraging a nonlinear multidimensional estimation of heritability, we provide evidence that individual variability in the size and topographic organization of cortical networks are under genetic control. Using twin and family data from the Human Connectome Project (n = 1,023), we find increased variability and reduced heritability in the size of heteromodal association networks (h2 : M = 0.34, SD = 0.070), relative to unimodal sensory/motor cortex (h2 : M = 0.40, SD = 0.097). We then demonstrate that the spatial layout of cortical networks is influenced by genetics, using our multidimensional estimation of heritability (h2-multi; M = 0.14, SD = 0.015). However, topographic heritability did not differ between heteromodal and unimodal networks. Genetic factors had a regionally variable influence on brain organization, such that the heritability of network topography was greatest in prefrontal, precuneus, and posterior parietal cortex. Taken together, these data are consistent with relaxed genetic control of association cortices relative to primary sensory/motor regions and have implications for understanding population-level variability in brain functioning, guiding both individualized prediction and the interpretation of analyses that integrate genetics and neuroimaging.

A Structure-Function Substrate of Memory for Spatial Configurations in Medial and Lateral Temporal Cortices

Prior research has shown a role of the medial temporal lobe, particularly the hippocampal-parahippocampal complex, in spatial cognition. Here, we developed a new paradigm, the conformational shift spatial task (CSST), which examines the ability to encode and retrieve spatial relations between unrelated items. This task is short, uses symbolic cues, incorporates two difficulty levels, and can be administered inside the scanner. A cohort of 48 healthy young adults underwent the CSST, together with a set of behavioral measures and multimodal magnetic resonance imaging (MRI). Inter-individual differences in CSST performance correlated with scores on an established spatial memory paradigm, but neither with episodic memory nor mnemonic discrimination, supporting specificity. Analyzing high-resolution structural MRI data, individuals with better spatial memory showed thicker medial and lateral temporal cortices. Functional relevance of these findings was supported by task-based functional MRI analysis in the same participants and ad hoc meta-analysis. Exploratory resting-state functional MRI analyses centered on clusters of morphological effects revealed additional modulation of intrinsic network integration, particularly between lateral and medial temporal structures. Our work presents a novel spatial memory paradigm and supports an integrated structure-function substrate in the human temporal lobe. Task paradigms are programmed in python and made open access.

Deep learning applications for the classification of psychiatric disorders using neuroimaging data: Systematic review and meta-analysis

Deep learning (DL) methods have been increasingly applied to neuroimaging data to identify patients with psychiatric and neurological disorders. This review provides an overview of the different DL applications within psychiatry and compares DL model accuracy to standard machine learning (SML). Fifty-three articles were included for qualitative analysis, primarily investigating autism spectrum disorder (ASD; n = 22), schizophrenia (SZ; n = 22) and attention-deficit/hyperactivity disorder (ADHD; n = 9). Thirty-two of the thirty-five studies that directly compared DL to SML reported a higher accuracy for DL. Only sixteen studies could be included in a meta-regression to quantitatively compare DL and SML performance. This showed a higher odds ratio for DL models, though the comparison attained significance only for ASD. Our results suggest that deep learning of neuroimaging data is a promising tool for the classification of individual psychiatric patients. However, it is not yet used to its full potential: most studies use pre-engineered features, whereas one of the main advantages of DL is its ability to learn representations of minimally processed data. Our current evaluation is limited by minimal reporting of performance measures to enable quantitative comparisons, and the restriction to ADHD, SZ and ASD as current research focusses on large publicly available datasets. To truly uncover the added value of DL, we need carefully designed comparisons of SML and DL models which are yet rarely performed.

Artificial Intelligence for Mental Health Care: Clinical Applications, Barriers, Facilitators, and Artificial Wisdom

Artificial intelligence (AI) is increasingly employed in health care fields such as oncology, radiology, and dermatology. However, the use of AI in mental health care and neurobiological research has been modest. Given the high morbidity and mortality in people with psychiatric disorders, coupled with a worsening shortage of mental health care providers, there is an urgent need for AI to help identify high-risk individuals and provide interventions to prevent and treat mental illnesses. While published research on AI in neuropsychiatry is rather limited, there is a growing number of successful examples of AI's use with electronic health records, brain imaging, sensor-based monitoring systems, and social media platforms to predict, classify, or subgroup mental illnesses as well as problems such as suicidality. This article is the product of a study group held at the American College of Neuropsychopharmacology conference in 2019. It provides an overview of AI approaches in mental health care, seeking to help with clinical diagnosis, prognosis, and treatment, as well as clinical and technological challenges, focusing on multiple illustrative publications. Although AI could help redefine mental illnesses more objectively, identify them at a prodromal stage, personalize treatments, and empower patients in their own care, it must address issues of bias, privacy, transparency, and other ethical concerns. These aspirations reflect human wisdom, which is more strongly associated than intelligence with individual and societal well-being. Thus, the future AI or artificial wisdom could provide technology that enables more compassionate and ethically sound care to diverse groups of people.

Accurate brain age prediction with lightweight deep neural networks

Deep learning has huge potential for accurate disease prediction with neuroimaging data, but the prediction performance is often limited by training-dataset size and computing memory requirements. To address this, we propose a deep convolutional neural network model, Simple Fully Convolutional Network (SFCN), for accurate prediction of brain age using T1-weighted structural MRI data. Compared with other popular deep network architectures, SFCN has fewer parameters, so is more compatible with small dataset size and 3D volume data. The network architecture was combined with several techniques for boosting performance, including data augmentation, pre-training, model regularization, model ensemble and prediction bias correction. We compared our overall SFCN approach with several widely-used machine learning models. It achieved state-of-the-art performance in UK Biobank data (N = 14,503), with mean absolute error (MAE) = 2.14y in brain age prediction and 99.5% in sex classification. SFCN also won (both parts of) the 2019 Predictive Analysis Challenge for brain age prediction, involving 79 competing teams (N = 2,638, MAE = 2.90y). We describe here the details of our approach, and its optimisation and validation. Our approach can easily be generalised to other tasks using different image modalities, and is released on GitHub.