More than the sum of its parts: Merging network psychometrics and network neuroscience with application in autism

Network approaches that investigate the interaction between symptoms or behaviours have opened new ways of understanding psychological phenomena in health and disorder. In parallel, network approaches that characterise the interaction between brain regions have become the dominant approach in neuroimaging research. Combining these parallel approaches would enable new insights into the interaction between behaviours and their brain-level correlates. In this paper, we introduce a methodology for combining network psychometrics and network neuroscience. This approach utilises the information from the psychometric network to obtain neural correlates for each node in the psychometric network (network-based regression). We illustrate the approach by highlighting the interaction between autistic traits and their resting-state functional associations. To this end, we utilise data from 172 male autistic participants (10–21 years) from the autism brain data exchange (ABIDE, ABIDE-II). Our results indicate that the network-based regression approach can uncover both unique and shared neural correlates of behavioural measures. In addition, the methodology enables us to isolate mechanisms at the brain-level that are unique to particular behavioural variables. For instance, our example analysis indicates that the overlap between communication and social difficulties is not reflected in the overlap between their functional correlates.

Towards Biophysical Markers of Depression Vulnerability

A major difficulty with treating psychiatric disorders is their heterogeneity: different neural causes can lead to the same phenotype. To address this, we propose describing the underlying pathophysiology in terms of interpretable, biophysical parameters of a neural model derived from the electroencephalogram. We analyzed data from a small patient cohort of patients with depression and controls. We constructed biophysical models that describe neural dynamics in a cortical network activated during a task that is used to assess depression state. We show that biophysical model parameters are biomarkers, that is, variables that allow subtyping of depression at a biological level. They yield a low dimensional, interpretable feature space that allowed description of differences between individual patients with depressive symptoms. They capture internal heterogeneity/variance of depression state and achieve significantly better classification than commonly used EEG features. Our work is a proof of concept that a combination of biophysical models and machine learning may outperform earlier approaches based on classical statistics and raw brain data.

Altered brain criticality in Schizophrenia: New insights from MEG

Schizophrenia has a complex etiology and symptomatology that is difficult to untangle. After decades of research, important advancements towards a central biomarker are still lacking. One of the missing pieces is a better understanding of how non-linear neural dynamics are altered in this patient population. In this study, the resting-state neuromagnetic signals of schizophrenia patients and healthy controls were analyzed in the framework of criticality. When biological systems like the brain are in a state of criticality, they are thought to be functioning at maximum efficiency (e.g., optimal communication and storage of information) and with maximum adaptability to incoming information. Here, we assessed the self-similarity and multifractality of resting-state brain signals recorded with magnetoencephalography in patients with schizophrenia patients and in matched controls. Our analysis showed a clear ascending, rostral to caudal gradient of self-similarity values in healthy controls, and an opposite gradient for multifractality (descending values, rostral to caudal). Schizophrenia patients had similar, although attenuated, gradients of self-similarity and multifractality values. Statistical tests showed that patients had higher values of self-similarity than controls in fronto-temporal regions, indicative of more regularity and memory in the signal. In contrast, patients had less multifractality than controls in the parietal and occipital regions, indicative of less diverse singularities and reduced variability in the signal. In addition, supervised machine-learning, based on logistic regression, successfully discriminated the two groups using measures of self-similarity and multifractality as features. Our results provide new insights into the baseline cognitive functioning of schizophrenia patients by identifying key alterations of criticality properties in their resting-state brain data.

Compositional Restricted Boltzmann Machines Unveil the Brain-Wide Organization of Neural Assemblies

Patterns of endogenous activity in the brain reflect a stochastic exploration of the neuronal state space that is constrained by the underlying assembly organization of neurons. Yet it remains to be shown that this interplay between neurons and their assembly dynamics indeed suffices to generate whole-brain data statistics. Here we recorded the activity from ~40,000 neurons simultaneously in zebrafish larvae, and show that a data-driven network model of neuron-assembly interactions can accurately reproduce the mean activity and pairwise correlation statistics of their spontaneous activity. This model, the compositional Restricted Boltzmann Machine, unveils ~200 neural assemblies, which compose neurophysiological circuits and whose various combinations form successive brain states. From this, we mathematically derived an interregional functional connectivity matrix, which is conserved across individual animals and correlates well with structural connectivity. This novel, assembly-based generative model of brain-wide neural dynamics enables physiology-bound perturbation experiments in silico.

Mitochondrial-nuclear cross-talk in the human brain is modulated by cell type and perturbed in neurodegenerative disease

Mitochondrial dysfunction contributes to the pathogenesis of many neurodegenerative diseases. The mitochondrial genome encodes core respiratory chain proteins, but the vast majority of mitochondrial proteins are nuclear-encoded, making interactions between the two genomes vital for cell function. Here, we examine these relationships by comparing mitochondrial and nuclear gene expression across different regions of the human brain in healthy and disease cohorts. We find strong regional patterns that are modulated by cell-type and reflect functional specialisation. Nuclear genes causally implicated in sporadic Parkinson’s and Alzheimer’s disease (AD) show much stronger relationships with the mitochondrial genome than expected by chance, and mitochondrial-nuclear relationships are highly perturbed in AD cases, particularly through synaptic and lysosomal pathways, potentially implicating the regulation of energy balance and removal of dysfunction mitochondria in the etiology or progression of the disease. Finally, we present MitoNuclearCOEXPlorer, a tool to interrogate key mitochondria-nuclear relationships in multi-dimensional brain data.

Brain Data Standards Ontology: A data-driven ontology of transcriptomically defined cell types in the primary motor cortex

Large scale single cell omics profiling is revolutionising our understanding of cell types, especially in complex organs like the brain. This presents both an opportunity and a challenge for cell ontologies. Annotation of cell types in single cell 'omics data typically uses unstructured free text, making comparison and mapping of annotation between datasets challenging. Annotation with cell ontologies is key to overcoming this challenge, but this will require meeting the challenge of extending cell ontologies representing classically defined cell types by defining and classifying cell types directly from data. Here we present the Brain Data Standards Ontology (BDSO), a data driven ontology that is built as an extension to the Cell Ontology (CL). It supports two major use cases: cell type annotation, and navigation, search, and organisation of a web application integrating single cell omics datasets for the mammalian primary motor cortex. The ontology is built using a semi-automated pipeline that interlinks cell type taxonomies and necessary and sufficient marker genes, and imports relevant ontology modules derived from external ontologies. Overall, the BDS ontology provides an underlying structure that supports these use cases, while remaining sustainable and extensible through automation as our knowledge of brain cell type expands.

Multispectral single chip reconstruction using DNNs with application to open neurosurgery

Multispectral imaging devices incorporating up to 256 different spectral channels have recently become available for various healthcare applications, as e.g. laparoscopy, gastroscopy, dermatology or perfusion imaging for wound analysis. Currently, the use of such devices is limited due to very high investment costs and slow capture times. To compensate these shortcomings, single sensors with spectral masking on the pixel level have been proposed. Hence, adequate spectral reconstruction methods are needed. Within this work, two deep convolutional neural networks (DCNN) architectures for spectral image reconstruction from single sensors are compared with each other. Training of the networks is based on a huge collection of different MSI imagestacks, which have been subsampled, simulating 16-channel single sensors with spectral masking. We define a training, validation and test set (‘HITgoC’) resulting in 351 training (631.128 sub-images), 99 validation (163.272 sub-images) and 51 test images. For the application in the field of neurosurgery an additional testing set of 36 image stacks from the Nimbus data collection is used, depicting MSI brain data during open surgery. Two DCNN architectures were compared to bilinear interpolation (BI) and an intensity difference (ID) algorithm. The DCNNs (ResNet-Shinoda) were trained on HITgoC and consist of a preprocessing step using BI or ID and a refinement part using a ResNet structure. Similarity measures used were PSNR, SSIM and MSE between predicted and reference images. We calculated the similarity measures for HitgoC and Nimbus data and determined differences of the mean similarity measure values achieved with the ResNet-ID and baseline algorithms such as BI algorithm and ResNet-Shinoda. The proposed method achieved better results against BI in SSIM (.0644 vs. .0252), PSNR (15.3 dB vs. 9.1 dB) and 1-MSE*100 (.0855 vs. .0273) and compared to ResNet-Shinoda in SSIM (.0103 vs. .0074), PSNR (3.8 dB vs. 3.6 dB) and 1-MSE*100 (.0075 vs. .0047) for HITgoC/Nimbus. In this study, significantly better results for spectral reconstruction in MSI images of open neurosurgery was achieved using a combination of ID-interpolation and ResNet structure compared to standard methods.