Sparse deep neural networks on imaging genetics for schizophrenia case-control classification

Multiview hyperedge-aware hypergraph embedding learning for multisite, multiatlas fMRI based functional connectivity network analysis

Recently, functional magnetic resonance imaging (fMRI) based functional connectivity network (FCN) analysis via graph convolutional networks (GCNs) has shown promise for automated diagnosis of brain diseases by regarding the FCNs as irregular graph-structured data. However, multiview information and site influences of the FCNs in a multisite, multiatlas fMRI scenario have been understudied. In this paper, we propose a Class-consistency and Site-independence Multiview Hyperedge-Aware HyperGraph Embedding Learning (CcSi-MHAHGEL) framework to integrate FCNs constructed on multiple brain atlases in a multisite fMRI study. Specifically, for each subject, we first model brain network as a hypergraph for every brain atlas to characterize high-order relations among multiple vertexes, and then introduce a multiview hyperedge-aware hypergraph convolutional network (HGCN) to extract a multiatlas-based FCN embedding where hyperedge weights are adaptively learned rather than employing the fixed weights precalculated in traditional HGCNs. In addition, we formulate two modules to jointly learn the multiatlas-based FCN embeddings by considering the between-subject associations across classes and sites, respectively, i.e., a class-consistency module to encourage both compactness within every class and separation between classes for promoting discrimination in the embedding space, and a site-independence module to minimize the site dependence of the embeddings for mitigating undesired site influences due to differences in scanning platforms and/or protocols at multiple sites. Finally, the multiatlas-based FCN embeddings are fed into a few fully connected layers followed by the soft-max classifier for diagnosis decision. Extensive experiments on the ABIDE demonstrate the effectiveness of our method for autism spectrum disorder (ASD) identification. Furthermore, our method is interpretable by revealing ASD-relevant brain regions that are biologically significant.

Accounting for temporal variability in functional magnetic resonance imaging improves prediction of intelligence

Neuroimaging-based prediction methods for intelligence have seen a rapid development. Among different neuroimaging modalities, prediction using functional connectivity (FC) has shown great promise. Most literature has focused on prediction using static FC, with limited investigations on the merits of such analysis compared to prediction using dynamic FC or region-level functional magnetic resonance imaging (fMRI) times series that encode temporal variability. To account for the temporal dynamics in fMRI, we propose a bi-directional long short-term memory (bi-LSTM) approach that incorporates feature selection mechanism. The proposed pipeline is implemented via an efficient algorithm and applied for predicting intelligence using region-level time series and dynamic FC. We compare the prediction performance using different fMRI features acquired from the Adolescent Brain Cognitive Development (ABCD) study involving nearly 7000 individuals. Our detailed analysis illustrates the consistently inferior performance of static FC compared to region-level time series or dynamic FC for single and combined rest and task fMRI experiments. The joint analysis of task and rest fMRI leads to improved intelligence prediction under all models compared to using fMRI from only one experiment. In addition, the proposed bi-LSTM pipeline based on region-level time series identifies several shared and differential important brain regions across fMRI experiments that drive intelligence prediction. A test-retest analysis of the selected regions shows strong reliability across cross-validation folds. Given the large sample size of ABCD study, our results provide strong evidence that superior prediction of intelligence can be achieved by accounting for temporal variations in fMRI.

Sampling inequalities affect generalization of neuroimaging-based diagnostic classifiers in psychiatry

BACKGROUND

The development of machine learning models for aiding in the diagnosis of mental disorder is recognized as a significant breakthrough in the field of psychiatry. However, clinical practice of such models remains a challenge, with poor generalizability being a major limitation.

METHODS

Here, we conducted a pre-registered meta-research assessment on neuroimaging-based models in the psychiatric literature, quantitatively examining global and regional sampling issues over recent decades, from a view that has been relatively underexplored. A total of 476 studies (n = 118,137) were included in the current assessment. Based on these findings, we built a comprehensive 5-star rating system to quantitatively evaluate the quality of existing machine learning models for psychiatric diagnoses.

RESULTS

A global sampling inequality in these models was revealed quantitatively (sampling Gini coefficient (G) = 0.81, p < .01), varying across different countries (regions) (e.g., China, G = 0.47; the USA, G = 0.58; Germany, G = 0.78; the UK, G = 0.87). Furthermore, the severity of this sampling inequality was significantly predicted by national economic levels (β =  - 2.75, p < .001, R²_adj = 0.40; r =  - .84, 95% CI: - .41 to - .97), and was plausibly predictable for model performance, with higher sampling inequality for reporting higher classification accuracy. Further analyses showed that lack of independent testing (84.24% of models, 95% CI: 81.0-87.5%), improper cross-validation (51.68% of models, 95% CI: 47.2-56.2%), and poor technical transparency (87.8% of models, 95% CI: 84.9-90.8%)/availability (80.88% of models, 95% CI: 77.3-84.4%) are prevailing in current diagnostic classifiers despite improvements over time. Relating to these observations, model performances were found decreased in studies with independent cross-country sampling validations (all p < .001, BF₁₀ > 15). In light of this, we proposed a purpose-built quantitative assessment checklist, which demonstrated that the overall ratings of these models increased by publication year but were negatively associated with model performance.

CONCLUSIONS

Together, improving sampling economic equality and hence the quality of machine learning models may be a crucial facet to plausibly translating neuroimaging-based diagnostic classifiers into clinical practice.

Imputed Gene Expression versus Single Nucleotide Polymorphism in Predicting Gray Matter Phenotypes

Genetics plays an important role in psychiatric disorders. A clinically relevant question is whether we can predict psychiatric traits from genetics, which holds promise for early detection and tailored intervention. Imputed gene expression, also known as genetically-regulated expression (GRE), reflects the tissue-specific regulatory effects of multiple single nucleotide polymorphisms (SNPs) on genes. In this work, we explored the utility of GRE for trait association studies and how the GRE-based polygenic risk score (gPRS) compared with SNP-based PRS (sPRS) in predicting psychiatric traits. A total of 13 schizophrenia-related gray matter networks identified in another study served as the target brain phenotypes for assessing genetic associations and prediction accuracies in 34,149 individuals from the UK Biobank cohort. GRE was computed leveraging MetaXcan and GTEx tools for 56,348 genes across 13 available brain tissues. We then estimated the effects of individual SNPs and genes separately on each tested brain phenotype in the training set. The effect sizes were then used to compute gPRS and sPRS in the testing set, whose correlations with the brain phenotypes were used to assess the prediction accuracy. The results showed that, with the testing sample size set to 1,138, for training sample sizes from 1,138 up to 33,011, overall both gPRS and sPRS successfully predicted the brain phenotypes with significant correlations observed in the testing set, and higher accuracies noted for larger training sets. In addition, gPRS outperformed sPRS by showing significantly higher prediction accuracies across 13 brain phenotypes, with greater improvement noted for training sample sizes below ∼15,000. These findings support that GRE may serve as the primary genetic variable in brain phenotype association and prediction studies. Future imaging genetic studies may consider GRE as an option depending on the available sample size.

A deep residual model for characterization of 5D spatiotemporal network dynamics reveals widespread spatiodynamic changes in schizophrenia

Schizophrenia is a severe brain disorder with serious symptoms including delusions, disorganized speech, and hallucinations that can have a long-term detrimental impact on different aspects of a patient's life. It is still unclear what the main cause of schizophrenia is, but a combination of altered brain connectivity and structure may play a role. Neuroimaging data has been useful in characterizing schizophrenia, but there has been very little work focused on voxel-wise changes in multiple brain networks over time, despite evidence that functional networks exhibit complex spatiotemporal changes over time within individual subjects. Recent studies have primarily focused on static (average) features of functional data or on temporal variations between fixed networks; however, such approaches are not able to capture multiple overlapping networks which change at the voxel level. In this work, we employ a deep residual convolutional neural network (CNN) model to extract 53 different spatiotemporal networks each of which captures dynamism within various domains including subcortical, cerebellar, visual, sensori-motor, auditory, cognitive control, and default mode. We apply this approach to study spatiotemporal brain dynamism at the voxel level within multiple functional networks extracted from a large functional magnetic resonance imaging (fMRI) dataset of individuals with schizophrenia (N = 708) and controls (N = 510). Our analysis reveals widespread group level differences across multiple networks and spatiotemporal features including voxel-wise variability, magnitude, and temporal functional network connectivity in widespread regions expected to be impacted by the disorder. We compare with typical average spatial amplitude and show highly structured and neuroanatomically relevant results are missed if one does not consider the voxel-wise spatial dynamics. Importantly, our approach can summarize static, temporal dynamic, spatial dynamic, and spatiotemporal dynamics features, thus proving a powerful approach to unify and compare these various perspectives. In sum, we show the proposed approach highlights the importance of accounting for both temporal and spatial dynamism in whole brain neuroimaging data generally, shows a high-level of sensitivity to schizophrenia highlighting global but spatially unique dynamics showing group differences, and may be especially important in studies focused on the development of brain-based biomarkers.

Consistent brain structural abnormalities and multisite individualised classification of schizophrenia using deep neural networks

BACKGROUND

Previous analyses of grey and white matter volumes have reported that schizophrenia is associated with structural changes. Deep learning is a data-driven approach that can capture highly compact hierarchical non-linear relationships among high-dimensional features, and therefore can facilitate the development of clinical tools for making a more accurate and earlier diagnosis of schizophrenia.

AIMS

To identify consistent grey matter abnormalities in patients with schizophrenia, 662 people with schizophrenia and 613 healthy controls were recruited from eight centres across China, and the data from these independent sites were used to validate deep-learning classifiers.

METHOD

We used a prospective image-based meta-analysis of whole-brain voxel-based morphometry. We also automatically differentiated patients with schizophrenia from healthy controls using combined grey matter, white matter and cerebrospinal fluid volumetric features, incorporated a deep neural network approach on an individual basis, and tested the generalisability of the classification models using independent validation sites.

RESULTS

We found that statistically reliable schizophrenia-related grey matter abnormalities primarily occurred in regions that included the superior temporal gyrus extending to the temporal pole, insular cortex, orbital and middle frontal cortices, middle cingulum and thalamus. Evaluated using leave-one-site-out cross-validation, the performance of the classification of schizophrenia achieved by our findings from eight independent research sites were: accuracy, 77.19-85.74%; sensitivity, 75.31-89.29% and area under the receiver operating characteristic curve, 0.797-0.909.

CONCLUSIONS

These results suggest that, by using deep-learning techniques, multidimensional neuroanatomical changes in schizophrenia are capable of robustly discriminating patients with schizophrenia from healthy controls, findings which could facilitate clinical diagnosis and treatment in schizophrenia.

A novel generation adversarial network framework with characteristics aggregation and diffusion for brain disease classification and feature selection

Imaging genetics provides unique insights into the pathological studies of complex brain diseases by integrating the characteristics of multi-level medical data. However, most current imaging genetics research performs incomplete data fusion. Also, there is a lack of effective deep learning methods to analyze neuroimaging and genetic data jointly. Therefore, this paper first constructs the brain region-gene networks to intuitively represent the association pattern of pathogenetic factors. Second, a novel feature information aggregation model is constructed to accurately describe the information aggregation process among brain region nodes and gene nodes. Finally, a deep learning method called feature information aggregation and diffusion generative adversarial network (FIAD-GAN) is proposed to efficiently classify samples and select features. We focus on improving the generator with the proposed convolution and deconvolution operations, with which the interpretability of the deep learning framework has been dramatically improved. The experimental results indicate that FIAD-GAN can not only achieve superior results in various disease classification tasks but also extract brain regions and genes closely related to AD. This work provides a novel method for intelligent clinical decisions. The relevant biomedical discoveries provide a reliable reference and technical basis for the clinical diagnosis, treatment and pathological analysis of disease.

Deep learning in neuroimaging data analysis: Applications, challenges, and solutions

Methods for the analysis of neuroimaging data have advanced significantly since the beginning of neuroscience as a scientific discipline. Today, sophisticated statistical procedures allow us to examine complex multivariate patterns, however most of them are still constrained by assuming inherent linearity of neural processes. Here, we discuss a group of machine learning methods, called deep learning, which have drawn much attention in and outside the field of neuroscience in recent years and hold the potential to surpass the mentioned limitations. Firstly, we describe and explain the essential concepts in deep learning: the structure and the computational operations that allow deep models to learn. After that, we move to the most common applications of deep learning in neuroimaging data analysis: prediction of outcome, interpretation of internal representations, generation of synthetic data and segmentation. In the next section we present issues that deep learning poses, which concerns multidimensionality and multimodality of data, overfitting and computational cost, and propose possible solutions. Lastly, we discuss the current reach of DL usage in all the common applications in neuroimaging data analysis, where we consider the promise of multimodality, capability of processing raw data, and advanced visualization strategies. We identify research gaps, such as focusing on a limited number of criterion variables and the lack of a well-defined strategy for choosing architecture and hyperparameters. Furthermore, we talk about the possibility of conducting research with constructs that have been ignored so far or/and moving toward frameworks, such as RDoC, the potential of transfer learning and generation of synthetic data.

Eye Movement Patterns Can Distinguish Schizophrenia From the Major Affective Disorders and Healthy Control Subjects

Background and hypothesis

No objective tests are currently available to help diagnosis of major psychiatric disorders. This study evaluates the potential of eye movement behavior patterns to predict schizophrenia subjects compared to those with major affective disorders and control groups.

Study design

Eye movements were recorded from a training set of UK subjects with schizophrenia (SCZ; n = 120), bipolar affective disorder (BPAD; n = 141), major depressive disorder (MDD; n = 136), and healthy controls (CON; n = 142), and from a hold-out set of 133 individuals with proportional group sizes. A German cohort of SCZ (n = 60) and a Scottish cohort of CON subjects (n = 184) acted as a second semi-independent test set. All patients met DSMIV and ICD10 criteria for SCZ, BPAD, and MDD. Data from 98 eye movement features were extracted. We employed a gradient boosted (GB) decision tree multiclass classifier to develop a predictive model. We calculated the area under the curve (AUC) as the primary performance metric.

Study results

Estimates of AUC in one-versus-all comparisons were: SCZ (0.85), BPAD (0.78), MDD (0.76), and CON (0.85). Estimates on part-external validation were SCZ (0.89) and CON (0.65). In all cases, there was good specificity but only moderate sensitivity. The best individual discriminators included free viewing, fixation duration, and smooth pursuit tasks. The findings appear robust to potential confounders such as age, sex, medication, or mental state at the time of testing.

Conclusions

Eye movement patterns can discriminate schizophrenia from major mood disorders and control subjects with around 80% predictive accuracy.

Recent advances in understanding neural correlates of anxiety disorders in children and adolescents

PURPOSE OF REVIEW

Anxiety disorders are some of the most common psychiatric diagnoses in children and adolescents, but attempts to improve outcome prediction and treatment have stalled. This review highlights recent findings on neural indices related to fear and anxiety that provide novel directions for attempts to create such improvements.

RECENT FINDINGS

Stimuli capable of provoking fear engage many brain regions, including the amygdala, medial prefrontal cortex, hippocampus, and bed nucleus of the stria terminalis. Studies in rodents suggest that sustained, low-level threats are particularly likely to engage the bed nucleus of the stria terminalis, which appears to malfunction in anxiety disorders. However, anxiety disorders, like most mental illnesses, appear less likely to arise from alterations in isolated brain regions than in distributed brain circuitry. Findings from large-scale studies of brain connectivity may reveal signs of such broadly distributed dysfunction, though available studies report small effect sizes. Finally, we review novel approaches with promise for using such large-scale data to detect clinically relevant, broadly distributed circuitry dysfunction.

SUMMARY

Recent work maps neural circuitry related to fear and anxiety. This circuitry may malfunction in anxiety disorders. Integrating findings from animal studies, big datasets, and novel analytical approaches may generate clinically relevant insights based on this recent work.

Sparse deep neural networks on imaging genetics for schizophrenia case-control classification

Deep learning methods hold strong promise for identifying biomarkers for clinical application. However, current approaches for psychiatric classification or prediction do not allow direct interpretation of original features. In the present study, we introduce a sparse deep neural network (DNN) approach to identify sparse and interpretable features for schizophrenia (SZ) case–control classification. An L₀‐norm regularization is implemented on the input layer of the network for sparse feature selection, which can later be interpreted based on importance weights. We applied the proposed approach on a large multi‐study cohort with gray matter volume (GMV) and single nucleotide polymorphism (SNP) data for SZ classification. A total of 634 individuals served as training samples, and the classification model was evaluated for generalizability on three independent datasets of different scanning protocols (N = 394, 255, and 160, respectively). We examined the classification power of pure GMV features, as well as combined GMV and SNP features. Empirical experiments demonstrated that sparse DNN slightly outperformed independent component analysis + support vector machine (ICA + SVM) framework, and more effectively fused GMV and SNP features for SZ discrimination, with an average error rate of 28.98% on external data. The importance weights suggested that the DNN model prioritized to select frontal and superior temporal gyrus for SZ classification with high sparsity, with parietal regions further included with lower sparsity, echoing previous literature. The results validate the application of the proposed approach to SZ classification, and promise extended utility on other data modalities and traits which ultimately may result in clinically useful tools.