Neuroinformatics and Computational Modelling
Poster #101: Adrien Osakwe
Title: SpaTM: Topic Models for Inferring Spatially Informed Transcriptional Programs
Abstract: Spatial transcriptomics has revolutionized our ability to characterize tissues and diseases by contextualizing gene expression with spatial organization. Available methods require researchers to either train a model using histology-based annotations or use annotation-free clustering approaches to uncover spatial domains. However, few methods provide researchers with a way to jointly analyze spatial data from both annotation-free and annotation-guided perspectives using consistent inductive biases and levels of interpretability. A single framework with consistent inductive biases ensures coherence and transferability across tasks, reducing the risks of conflicting assumptions. To this end, we propose the Spatial Topic Model (SpaTM), a topic-modeling framework capable of annotation-guided and annotation-free analysis of spatial transcriptomics data. SpaTM can be used to learn gene programs that represent histology-based annotations while providing researchers with the ability to infer spatial domains with an annotation-free approach if manual annotations are limited or noisy. We demonstrate SpaTM’s interpretability with its use of topic mixtures to represent cell states and transcriptional programs and how its intuitive framework facilitates the integration of annotation-guided and annotation-free analyses of spatial data with downstream analyses such as cell type deconvolution. Finally, we demonstrate how both approaches can be used to extend the analysis of large-scale snRNA-seq atlases with the inference of cell proximity and spatial annotations in human brains with Major Depressive Disorder.
Poster #102: Aliza Brzezinski-Rittner
Title: The mediating effect of white matter hyperintensities on spontaneous brain activity during aging
Abstract: Aging is associated with alterations in brain functional activity observed through resting-state functional Magnetic Resonance Imaging (rs-fMRI). White matter hyperintensities (WMH), indicative of cerebral small vessel disease, are prevalent in older adults. We examined whether WMH-burden mediates the relationship between healthy aging and intrinsic neural activity as measured by blood-oxygen-level-dependent signal. MRI data from 557 participants (51.7% female, 18-87 years), were obtained from the CamCAN dataset. Total and lobar WMH volumes were quantified from structural MRI, while regional fractional amplitude of low-frequency fluctuations (fALFF) and Regional Homogeneity (ReHo) metrics were derived from rs-fMRI. Correlation analyses assessed the relationships between age, WMH, and functional metrics. Mediation analyses were performed to investigate whether WMH-burden mediated the effect of age on spontaneous neural activity. fALFF exhibited significant (pfdr < 0.05) negative correlations with age and WMHs across the cortex; ReHo showed region-specific correlations: positive in frontal and occipital regions and negative in the temporal pole. Mediation analysis revealed significant effects of global-WMH on the impact of age on fALFF, predominantly in occipital and temporal regions. The mediation pattern of frontal-WMH-burden mirrored that of global-WMHs. Parietal and temporal-WMHs demonstrated significant but weaker mediation effects. Occipital-WMHs showed localized and modest effects. WMHs did not significantly mediate the age-related changes in regional ReHo.The impact of aging on spontaneous neural activity, specifically fALFF, is mediated by WMH-burden. The mediation effect is pronounced for frontal-WMHs, often attributed to vascular origin, suggesting a potential link between vascular processes and age-related changes in neural activity.
Poster #103: Anisleidy Gonzalez-Mitjans
Title: AI-Driven Neurodiagnostics: A Scalable Framework for EEG Anomaly Detection Using a Distributed-Delay Neural Mass Model
Abstract: The integration of biophysically grounded neural simulations with Artificial Intelligence (AI) has the potential to transform clinical neurodiagnostics by overcoming the inherent challenges of limited pathological EEG datasets. We present a novel AI-driven framework that leverages a Distributed-Delay Neural Mass Model (DD-NMM) to generate synthetic EEG signals replicating both healthy and pathological brain states. Through systematic parameter tuning and domain-specific data augmentation, we enrich the diversity of simulated signals, enabling robust anomaly detection using machine learning techniques. Our approach integrates supervised classification and unsupervised one-class anomaly detection, achieving over 95% accuracy in synthetic tests and over 89% when applied to real EEG data from epilepsy patients and healthy volunteers. By providing an engineered solution that bridges computational neuroscience with AI, this framework enhances early seizure detection, adaptive neurofeedback, and brain-computer interface applications. Our results demonstrate that theory-driven simulation, combined with state-of-the-art machine learning, can address critical gaps in medical AI, significantly advancing clinical neuroengineering.
Poster #104: Dane Malenfant
Title: Reciprocity in Spite of Inconsistent Time Preferences
Abstract: Humans and animals exhibit complex behaviours, especially with other humans and animals respectively. These behaviours can be cooperative like reciprocating of resources and span large distances. One such example exists within the Plains Indigenous nations who had a system of reciprocity involving effigies called Manitohkan with food, tools and medicine for travellers but had a cultural expectation of only taking what you need and giving away what you do not need. To test whether reinforcement learning agents can learn to develop a critical episodic memory of the cooperative action: giving away unneeded items. We created a novel multi-agent reinforcement learning (MARL) environment, Mācistan, based on the key-to-door (K2D) task to investigate reciprocity as a credit assignment problem. K2D requires agents to learn to exploit episodic memories associated with finding a key and opening a door to receive a reward signal. Mācistan additionally requires an agent to drop a key for another agent. If all doors are opened the agents both receive an additional larger reward signal. We empirically demonstrate nine different MARL algorithms either collapse or perform within 1% above chance except the policy gradient algorithm. We analyze the algorithm by extending the axioms of time preferences in multi-objective research and demonstrate a violation of dynamic consistency when another agent causes the collective reward to be unpredictable through time. We then derived a correction term to stabilize an agent’s “preferences” and retain an episodic memory of the giving away unneeded items.
Poster #105: Derek Newman
Title: EEG Criticality as a Prognostic Tool for Functional Outcomes in Sedated Pediatric Intensive Care Patients
Abstract: Predicting recovery in sedated pediatric patients is difficult due to the lack of reliable prognostic markers that don’t rely on behavior. EEG provides a real-time measure of brain activity that may improve prognostication. EEG criticality features reflect the brain’s balance between order and chaos, capturing its capacity for information processing and adaptability. These features such as Lempel-Ziv complexity, Hurst exponent, and spectral properties have been used to assess levels of consciousness in both adults and children. This study assessed whether EEG features during sedation predict recovery in sedated PICU patients. We hypothesized that EEG dynamics closer to criticality would indicate better outcomes. We retrospectively analyzed bedside EEG data from sedated PICU patients (n = 32). Participants ranged in age from 6 to 16 years (mean age = 11.34±3.19 years). Functional outcome was assessed using the Glasgow Outcome Scale – Extended (GOS-E) three months post-injury. Patients received Propofol, Midazolam, or Dexmedetomidine. Mann-Whitney U tests compared EEG features between the good (GOSE >= 4, n = 22) and poor recovery (GOSE < 4, n = 10) groups. Machine learning models were trained on 14 EEG features, etiology, sex, and age as predictors. Model performance was evaluated using stratified 5-fold cross-validation with bootstrap resampling to balance groups. Of the 14 EEG features analyzed, seven showed statistically significant differences between recovery groups. Logistic regression achieved 81% (±4.3%) accuracy in predicting patient recovery. EEG criticality features hold prognostic value in sedated PICU patients. Machine learning models using these features accurately predicted recovery, reinforcing their clinical utility.
Poster #106: Dylan Gharibian
Title: An Exploratory Analysis of Essential Tremor and Associated Phenotypes
Abstract: Essential Tremor (ET) is a highly heterogeneous movement disorder with a strong genetic basis. The etiology of ET however is unclear, largely due to its clinical heterogeneity. Within the UK Biobank, we noted a strong increase in total diagnosis counts in individuals with ET across sexes and age brackets. To establish a directional, causal relationship between ET and common cooccurring phenotypes, we conducted a Mendelian Randomization (MR ) experiment. Causality between traits could not be established however, which could be attributed to a high degree of genetic pleiotropy between ET and other phenotypes. Therefore, we employed genomic structural equation modelling to classify pleiotropic traits into latent variables to better characterize the observed pleiotropy. To then dissect the shared genetic landscape of ET, we examined the SNPs driving the pleiotropy of ET with cooccurring disorders. We therefore identified pleiotropic SNPs across traits, mapped them to genes, and then conducted gene ontology enrichment analyses of those genes to identify biochemical pathways relevant to specific ET-comorbidity associations.
Poster #107: Dylan Mann-Krzisnik (Trainee Flash Talks)
Title: Reciprocity in Spite of Inconsistent Time Preferences
Abstract: Humans and animals exhibit complex behaviours, especially with other humans and animals respectively. These behaviours can be cooperative like reciprocating of resources and span large distances. One such example exists within the Plains Indigenous nations who had a system of reciprocity involving effigies called Manitohkan with food, tools and medicine for travellers but had a cultural expectation of only taking what you need and giving away what you do not need. To test whether reinforcement learning agents can learn to develop a critical episodic memory of the cooperative action: giving away unneeded items. We created a novel multi-agent reinforcement learning (MARL) environment, Mācistan, based on the key-to-door (K2D) task to investigate reciprocity as a credit assignment problem. K2D requires agents to learn to exploit episodic memories associated with finding a key and opening a door to receive a reward signal. Mācistan additionally requires an agent to drop a key for another agent. If all doors are opened the agents both receive an additional larger reward signal. We empirically demonstrate nine different MARL algorithms either collapse or perform within 1% above chance except the policy gradient algorithm. We analyze the algorithm by extending the axioms of time preferences in multi-objective research and demonstrate a violation of dynamic consistency when another agent causes the collective reward to be unpredictable through time. We then derived a correction term to stabilize an agent’s “preferences” and retain an episodic memory of the giving away unneeded items.
Poster #108: Eric Ceballos
Title: Mapping neuropeptide signaling in the human brain
Abstract: Neuropeptides are functionally diverse signaling molecules in the brain, regulating a wide range of basal bodily and cognitive processes. Despite their importance, the distribution and function of neuropeptides in the human brain remains underexplored. Here we comprehensively map the organization of human whole-brain neuropeptide receptors across multiple levels of description; from molecular and cellular embedding to mesoscale connectivity and macroscale cognitive specialization. Using gene transcription as a proxy, we reconstruct a topographic cortical and subcortical atlas of neuropeptide receptors for 38 neuropeptide receptors, across 14 different neuropeptide families. We find that most neuropeptide receptors are highly expressed either in cortex or subcortex, delineating an anatomical cortical-subcortical gradient. Neuropeptides preferentially co-localize with metabotropic neurotransmitters, suggesting a system-wide correspondence between slow-acting molecular signaling mechanisms. Mapping neuropeptide receptors and their cognate ligands onto white-matter connectomes, we demonstrate that specific neuropeptides families shape electrophysiological and haemodynamic inter-regional connectivity. To investigate the behavioural consequences of distributed neuropeptide systems, we apply meta-analytic decoding to neuropeptide maps and show a gradient of functions, from sensory-cognitive to reward and bodily functions. Finally, evolutionary analysis indicates extended positive selection for neuropeptides in early mammals, suggesting that refinement of neuropeptides coincides with the emergence of neocortex and higher cognitive function. Collectively, these results show that the neuropeptide receptors are highly organized across the human brain and closely intertwined with multiple features of brain structure and function.
Poster #109: Haile Kassahun
Title: Design and Optimization of an Axial Gradient Coil for Low-field Halbach Array MRI scanners
Abstract: Magnetic resonance imaging (MRI) is a crucial imaging instrument for the visualization of the brain. Close to 80% of the world’s population does not have access to MRI for the discovery or management of neurological disease. High-field MRI systems with advanced gradient and RF coils are expensive and unaffordable for low- and middle-income countries. Therefore, a portable low-field MRI that is affordable and accessible for low- and middle-income countries is desired. Halbach array magnets are usually preferred for low-field, portable, point-of-care scanners due to their ability to generate a more homogeneous magnetic field. An efficient gradient coil design for Halbach array scanners is desired to get high-quality magnetic resonance (MR) images. However, there is still a challenge in obtaining desirable linearity and efficiency at the target diameter of spherical volume (DSV) for axial gradient coils using target field methods. In this work, we aim to investigate the discrete wire approach to design an efficient axial gradient coil for Halbach array scanners. The coil turns of each quadrant were parameterized using quasi-elliptic functions. The gradient coils are then optimized to maximize the coil’s efficiency while keeping the linearity error and the maximum field deviation less than 10% and lower than 5%, respectively. The coil geometric parameters, current, turn locations, center of a quadrant, and quasi-elliptic parameters are used as the design variables. Results showed that the designed Z (axial) gradient coils could achieve 1.21 mT/m/A over a cylindrical volume with a length of 6 cm and a diameter of 6 cm.
Poster #110: Jessica Royer
Title: Human cortical dynamics reflect graded contributions of local geometry and network topography
Abstract: The brain is a physically embedded and heavily interconnected system that expresses neural rhythms across multiple time scales. These resting dynamics result from the complex interplay of local and inter-regional factors, but the relative contribution of these mechanisms remains unclear. In this study, we explore geometric, microstructural, and connectome-level constraints on cortical neurophysiology measured with intracranial electroencephalography (iEEG). The MNI open iEEG atlas provides iEEG recordings acquired during conditions of resting wakefulness in 106 patients with intractable epilepsy. Providing broad cortical coverage, this dataset was used to derive an intrinsic coordinate system of brain dynamics. We first cross-correlated regional power spectrum densities and applied diffusion map embedding, a non-linear dimensionality reduction technique. The resulting eigenvectors (gradients) explaining the most variance in neural dynamic similarity recovered meaningful axes of cortical organization, separating motor and sensory regions, as well as unimodal and transmodal cortices. These patterns were mainly explainable by the brain’s geometric properties indexed by inter-regional geodesic distance (r=-0.376; pnull<0.001). However, multilinear models showed that dynamics in transmodal association regions were additionally explainable by microstructural similarity and connectivity. Indeed, implementing multimodal predictors selectively improved model accuracy in transmodal areas. Our findings were consistent when using data collected in the same individuals, suggesting subject-specificity and population-level generalizability. Together, these findings provide a normative reference space mapping of neocortical dynamics. Our results suggest that this space is constrained by gradual shifts in the balance of local and macroscale constraints, and highlight complex patterns of inter-regional cortical coordination in transmodal networks.
Poster #111: Jiadong Yan
Title: It is Possible to Prevent Mental Health Disorder by Regulating Cognition and Personality During Adolescence
Abstract: Adolescence is a sensitive window for the emergence of mental health disorder. Previous preventive interventions in adolescence have mainly focused on modifying risk exposure and enhancing positive coping skills. Due to the complex multifactorial nature and heterogeneity of mental health systems, these strategies showed limited effects. Meanwhile, although mounting evidence highlighted significant associations between behaviors, the possibility of preventing mental health disorder by regulating other behavior remains largely underexplored. Therefore, the current study employed an individualized multifactorial causality analysis to characterize mental health systems, investigate the causality linking other behavior with mental health, and further provide suggestions for novel personalized preventive strategies. In particular, using the ABCD data with comprehensive measures of brain structure, puberty, environment, and behavior, we first performed a cross-sectional multivariate mediation analysis to explore causal pathways linking cognition/personality and mental health. We then conducted a longitudinal interaction-inclusive model to investigate the previous causality effects. To explore personalized preventive strategies, we further performed the individual interpretive analysis and finally found 23 significant behavioral pairs with a preventive response rate exceeding 50%, among which regulating behavioral inhibition, negative urgency, and processing speed appeared most frequently as effective interventions, while psychosis symptoms and attention problems were the most commonly addressed outcomes. In conclusion, current study developed a novel analysis, addressing the multifactorial and individual nature of mental health system. We not only demonstrated the possibility of preventing mental health disorder by regulating cognition and personality during adolescence but also provided theoretical guidance for developing personalized interventions.
Poster #112: Mathieu Johnson
Title: 3D Topographical Distributions of Granule and Purkinje Cell Densities in Macaques and Humans
Abstract: The cerebellum contains over 50% of the brain’s neurons and is implicated in motor and cognitive functions, including motor coordination and planning, language, and emotion. Its contributions are mediated by cerebellar microcircuits reciprocally connected to specific cortical regions via the dentate nucleus, thalamus, and pons. At the cellular level, microcircuits are primarily composed of granule cells (integration of inputs from the cerebral cortex and brainstem) and Purkinje cells (output to the cerebellar nuclei). As previous approaches have generally focused on overall cell counts (e.g., entire cerebellum) or histology in specific regions, there is little information about the relative densities of cerebellar cortical cells across microcircuits and regions. Further, there is no information about white matter cell densities (oligodendrocytes/astrocytes) or species-specific differences. To explore potential relationships between cell types across different cerebellar regions, we serially acquired macaque cerebellum (Nissl stained, 398 slices, .343x.343×50μm in-plane), counted cells, and reconstructed cell counts in 3D. Preliminary results indicate a decrease in granule cell density within folia that may be linked to lower Purkinje cell densities in these regions. Our approach has been expanded to the human cerebellum (272 slices, 2x2x50μm in-plane), with the goal of producing detailed 3D maps of cell densities in grey and white matter for comparison. Given recent evolutionary expansion, we will also investigate if specific regions exhibit differential densities across the two species. This detailed cellular approach provides a deeper understanding of the structural organization of the cerebellum, helping further elucidate its function.
Poster #113: Mayumi Wong
Title: Comparative Analysis of Astrocyte Morphology Between Mouse and Marmoset
Abstract: Astrocytes are a major type of non-neuronal brain cells vital for neural circuit physiology. They are morphologically intricate and perform diverse functions through intracellular Ca2+ signalling. By closely interacting with synapses and vasculature, astrocytes modulate synaptic development, plasticity, blood flow and metabolic support in the neural circuit. As a result, astrocytes are heavily implicated in almost all neurological diseases. Despite their importance, the structure-function relationship of astrocytes remains poorly understood. Although recent studies have provided key insights into the rodent astrocytic nanoarchitecture (Salmon et al., Current Biology 2023), a nano-scale understanding of primate astrocytes is still lacking. In this study, we generated a marmoset dataset utilizing focused ion beam scanning electron microscopy (FIB-SEM). To facilitate large-scale analysis, a deep learning algorithm (3D-UNet) was developed to perform automatic segmentation of astrocytic features, significantly reducing manual effort. We established a novel electromicrograph segmentation pipeline using 3D-UNet coupled with manual correction of the false negatives, split errors and occasional false positives arising from membrane spill-over and erroneous astrocyte connections. Our comparative analysis revealed conserved structural features between mice and marmosets, such as a highly convoluted mesh of cellular processes and a scarcity of “donut” structures. Notable differences included variations in cytoplasm opacity and glycogen granule size, which could relate to cognitive differences and warrant further investigation. Future work will examine differences in organelle distribution and astrocyte connectivity across species. This study contributes to a deeper understanding of non-human primate astrocyte morphology and its potential implications for complex astrocyte functions.
Poster #114: Meaghan Smith
Title: Robustness of spatial correlations between gradients and intersubject correlations across movie-watching conditions
Abstract: Gradient analyses use dimensionality reduction to identify a macroscale organizational hierarchy from functional connectivity (FC). In contrast, intersubject correlations (ISCs) indicate how similarly a given brain region responds to the same stimulus across participants. Despite describing brain organization from different perspectives, both measures capture core features of functional architecture by delineating default mode regions from task-positive brain regions. Here, we leverage movie functional magnetic resonance imaging to generate movie gradients upon which intersubject synchronization is mapped in the form of ISCs. We compare these correlations across movies to investigate if the robustness of these relationships depends on movie content. Results demonstrated that at the whole-brain level, correlations between ISCs and the visual-to-default axis were high regardless of movie content, but they were most stable when computed across repeated viewing of the same clip. At the functional network level, the region with the strongest gradient-ISC correlations was the dorsal attention network. These results provide further support for a macroscale processing hierarchy within the brain that is exemplified under naturalistic conditions, while strong within-network correlations point to functional sub-hierarchies. Robust relationships across movies suggest that we might understand movie-watching as a brain state that is independent of movie-content. Overall, these findings suggest that when the brain is processing complex stimuli, there is a strong correspondence between FC patterns at a whole-brain level, and stimulus-evoked BOLD signal responses across subjects.
Poster #115: Mika Kaeja
Title: Latent Disconnection Networks Underlying Stroke-Induced Sensorimotor and Cognitive Deficits
Abstract: Stroke affects >108,000 Canadians annually, often resulting in persistent sensorimotor and cognitive deficits. These can arise from disruptions in the brain’s white matter (WM) network, where stroke-induced lesions disconnect functionally linked regions, impairing communication between areas critical for sensory, motor, and cognitive functions. However, behavioural tasks do not measure unitary constructs; a single task score may reflect multiple underlying behaviours with distinct neural substrates. This study applies a data-driven approach to decompose the relationship between WM disconnection and behavioural deficits in stroke. We used Partial Least Squares Singular Value Decomposition (PLS-SVD) to identify latent variables (LVs) capturing shared variance between voxel-wise disconnection maps and behavioural impairments, controlling for age and sex. MRI-derived lesion locations were used to compute individual WM disconnection maps in 24 stroke survivors. Patients completed four robotic exoskeleton tasks assessing sensorimotor and cognitive function: Ball on Bar (BOB), Object Hit (OHT), Arm Position Matching (APM), and Visually Guided Reach (VGR). Bootstrapping (1000 iterations) assessed LV stability. PLS-SVD identified four LVs linking task performance to distinct WM disconnection patterns, with LV1 explaining 94% of variance, followed by LV2 (2.5%), LV3 (2%), and LV4 (0.8%). LV1 was primarily driven by proprioceptive (APM) and motor (VGR) deficits, suggesting impairments in sensorimotor integration. Voxel weights for LV1 showed similar WM disconnection patterns to our prior univariate APM analysis. Ongoing behavioural decomposition aims to clarify whether finer-grained task components explain additional variance. This approach provides a novel framework for linking neuroanatomical disruptions to post-stroke behavioural deficits, refining neurobiological models of stroke recovery.
Poster #116: Miranda Medeiros
Title: Brain Imaging Phenotypes Associated with Polygenic Risk for Essential Tremor
Abstract: Background/Objectives: Essential tremor (ET) is a complex neurological disorder characterized by involuntary upper limb shaking. ET associated common variants can be leveraged to generate polygenic risk scores (PRS) to capture individuals’ ET specific genetic risk. Here, we aim to identify regions of the brain made vulnerable due to heightened genetic risk of ET in healthy subjects by investigating the association of white matter diffusion-weighted MRI (dMRI) , grey matter dMRI and morphometry with ET PRS in the UK Biobank (UKB) cohort. Methods: We paired genetic and imaging data from the UKB to model the associations of ET PRS on brain structure changes. Polygenic risk scores from healthy UKB participants were calculated using ET GWAS summary statistics (Liao et al., 2022) through PRS-CS. dMRI mean diffusion measures (DM) fractional anisotropy (FA), mean diffusivity (MD), and free-water (FW), were obtained across 73 anatomically curated tracts in the ORG fiber clustering white matter atlas (Zhang et al., 2018). We performed multiple regression analysis with glmnet (R) for variable selection and cross-validation and conducted a white matter to grey matter tract crossing and volume analysis. Results: Significant associations between ET PRS and dMRI cerebellar microstructure (e.g., cerebellar peduncles and intracerebellar Purkinje tracts) and associations between ET PRS and grey mater dMRI microstructure, in particular the red nucleus, hypothalamus, caudate, putamen, and several regions of a functional thalamus were identified. Conclusion: We have found that most regions associated with ET PRS are part of a known motor network involved in ET pathogenesis. Acknowledgments: We would like to thank the research participants and employees of 23andMe, Inc. for making this work possible.
Poster #117: Moohebat Pourmajidian
Title: Mapping Energy Metabolism Systems in the Human Brain
Abstract: Energy metabolism consists of a set of biological pathways that produce ATP using nutrients such as glucose and oxygen. These pathways also provide biomolecules critical for cellular growth and repair, making them integral to our understanding of brain structure and function. Despite extensive studies on brain glucose and oxygen uptake, the organization of downstream glucose metabolic pathways in the cortex remains largely unexplored. Here, we use whole-brain transcriptome data from the Allen Human Brain Atlas to study spatial cortical profiles of key energy pathways including glycolysis, pentose phosphate pathway, tricarboxylic acid cycle, oxidative phosphorylation and lactate metabolism. We create regional mean-expression maps of these pathways across the cortex using pathway-specific gene expression data. We show that energy pathways exhibit heterogenous cortical gene expression, with a dichotomy between primary ATP-producing pathways and the anabolic pentose phosphate pathway in the primary motor and sensory cortices, reflecting information processing hierarchy. These maps also exhibit unique relationships with the cellular and laminar organization of the cortex, pointing to higher energy demands of large pyramidal cells. Finally, we show that energy pathways exhibit unique developmental trajectories using a lifespan transcriptomics dataset of the human brain (42 donors, 19 female, ages 8 post conception weeks-40 years). The main ATP-producing pathways peak in childhood, scaling with cortical thickness, while the pentose phosphate pathway shows greater prenatal expression and declines in later life, mirroring brain tissue biosynthesis. Collectively, this study provides insight into the metabolic makeup of the human brain and its developmental demands.
Poster #118: Patricia Maidana Miguel (Trainee Flash Talks)
Title: Sex-specific developmental gene networks linking risk-taking behaviors in rodents and humans
Abstract: Early adversity increases the risk of psychopathology, including addictive-like behaviors, with emerging evidence of sex-specific effects. However, the underlying mechanisms remain unclear. This study investigated the effects of prenatal adversity on reward-seeking and risk-taking behaviors in rats, identifying sex-specific gene networks in the medial prefrontal cortex (mPFC) associated with these behaviors, that may serve as transcriptomic signatures in humans. Sprague Dawley dams were assigned to either a control (ad libitum diet) or food restriction (FR) group starting on gestational day 10. Offspring were tested at postnatal day 90 for lever-pressing behavior under a fixed-ratio 1 (FR1) reinforcement schedule and in the novelty-suppressed feeding (NSF) to assess reward-risk taking in an anxiogenic environment. Bulk RNA sequencing at P0, P21, and P90 (n=10/group) was analyzed using Weighted Gene Co-expression Network Analysis (WGCNA) and TimeNexus to identify FR-responsive gene subnetworks across development. Human orthologs from these subnetworks were used to derive sex-specific expression-based polygenic scores (ePRS), which were tested for associations with risk-taking behavior in the UK Biobank. FR offspring exhibited increased lever pressing for food reward (p=0.035, n=11-12/group), regardless of sex. However, only FR females showed faster initiation of eating in the NSF (p<0.05, Log-rank Mantel–Cox, n=16-19/group). In humans, the female ePRS was associated with risk-taking behavior in women (p=0.003, B=-0.03, n=208,995), while the male ePRS showed no effect in men (p=0.25, n=179,591). This study demonstrates a novel translational approach to identifying sex-specific brain gene networks that may serve as potential biomarkers of susceptibility to risk-taking behaviours in humans.
Poster #119: Shawniya Alageswaran
Title: Circuit mapping of V1 and M1 reveals layer-specific microcircuit structure
Abstract: The structure of microcircuits determines their function, which suggests that circuit mapping is a method of elucidating circuit function. The state-of-the-art mapping techniques like multiple patch clamp are challenging and are relatively low-throughput, making them unfeasible for mapping larger microcircuits. We therefore created optomapping, a high-throughput circuit mapping approach that relies on 2-photon optogenetics and patch-clamp electrophysiology. Using this approach, we investigated how microcircuits differ in primary visual (V1) and motor (M1) cortices, two sensory/input and motor/output areas, respectively. To express the soma-targeted opsin, ChroME, in neocortical pyramidal cells (PCs), we injected a viral cocktail into neonatal Emx1-Cre mice. In postnatal day 18-26 acute slices, we activated ChroME-expressing PCs using 1040-nm Ti:Sa-laser spiral scans. We patched PCs, basket cells (BCs), and Martinotti cells (MCs) in V1 and M1 while sequentially activating hundreds of surrounding PCs to find presynaptic PCs from all cortical layers. In both V1 and M1, the L5 PC to L5 BC pathway was the strongest, although it was stronger in V1. Lateral connectivity decayed faster for PC to PC than for PC to BC/MC synapses in V1, with preliminary data suggesting similar outcomes for M1. Excitatory input strengths onto PCs, BCs, and MCs were log-normally distributed in V1 and M1. The L4 PC to L2/3 PC pathway was denser (p < 0.001) and stronger (p<.01) in V1 than M1, which is fitting for a sensory/input area like V1. Thus, despite gross similarities across V1 and M1 microcircuits, optomapping reveals previously unseen layer-specific differences.
Poster #120: Tamires Marcal
Title: Characterization of Cognitive Structure Using Explainable Models and Multimodal Neuroimaging Maps
Abstract: Understanding the emergence of cognitive operations from the brain’s topographical organization is a fundamental goal in neuroscience. However, the roles and interactions of functional, structural, and chemical brain features in shaping cognitive structure have remained poorly characterized. This study aims to investigate these multimodal contributions to cognitive structure from a spatial patterning perspective. We utilized a comprehensive set of 48 brain maps from Neuromaps, encompassing functional MRI, structural MRI, PET, and ASL. To assess cognitive structure, we focused on CogPC1, a derivative component from Neurosynth, which represents the primary axis of variance in functional cognition. To examine the relationships between brain multimodal features, we developed machine learning models to predict CogPC1 – a general model using all modalities, along with four additional models, each based on a single modality. For results, the general model outperforms the unimodal models, explaining over 80% of the variance in CogPC1. In the general model, we found that functional connectivity in gradients 1, 7, and 6 had the highest contributions. A clear interaction effect with gradient 1 is visible on gradients 7 and 6 and norepinephrine transporter. The contribution of these features to CogPC1 varies according to the gradient 1 spectrum, affecting the slope, direction, and magnitude of their relationship with CogPC1. Our results reveal that functional connectivity gradients and density maps of receptors and transporters are key predictors of cognitive structure. Gradient 1 plays a crucial role in interacting with other brain features, suggesting that it encodes the operational regime of other brain features.
Poster #121: Yezhou Wang
Title: Capturing Cortical Functional Connectivity Shifts During Naturalistic Stimulation
Abstract: Cortical topographic organization underlies complex cognitive functions, yet how hierarchical network dynamics adapt to naturalistic stimuli remains unclear. Using high-resolution 7T MRI and regression dynamic causal modeling (rDCM), we investigated differences in sensory-driven hierarchical signal flow between resting and movie-watching states in 93 healthy adults. Effective connectivity (FC) analyses focused on the primary visual (V1) and auditory (A1) cortices, examining afferent and efferent connections across twelve defined brain networks. Movie watching enhanced FC between V1 and secondary visual/auditory networks, as well as between A1 and language/auditory networks (t>7.5, p<0.001). Efferent connections generally exhibited greater Movie-Rest Differences (MRD) than afferent ones, except in the frontoparietal network. To explore associations between MRD and visual-semantic features, we performed principal component analysis on motion energy (MEF) and semantic category features (SCF) of the movies. While MRD was not significantly related to MEF, it was significantly associated with SCF in the somatomotor, auditory, and default mode networks (p<0.05, FDR-corrected). These findings suggest that naturalistic stimuli enhance FC in sensory and language networks, reflecting the integration of multimodal information. Moreover, MRD was influenced by high-level semantic content rather than low-level perceptual features, highlighting the brain’s adaptation to cognitively rich environments. Our study provides a framework for understanding cortical FC reorganization during real-world cognition, offering insights into how sensory hierarchies facilitate multimodal processing.
Poster #122: Yigu Zhou
Title: Tracing the genetic gradients of the human brain
Abstract: Molecular pathways in the cerebral cortex interact with each other to specify the development of cortical structural and functional topography and stabilize adult patterns. We investigated the interaction between molecular axes by comparing spatial changes of cortical gene expression using microarray transcriptomic data of 7562 stable genes. First, we interpolated sparse microarray data to the dense fsaverage5 sampling mesh using ordinary kriging. Kriged gene expression patterns for selected neurotransmitter receptors consistently recapitulated group-averaged protein expression patterns via positron emissions tomography (PET; rPET(GABA_a)=.94, rPET(mGluR5)=.82, rPET(5HT1a)=.84, rPET(5HT2a)=.88), outperforming alternatives (K nearest neighbours: rPET(GABA_a)=.6, rPET(mGluR5)=.5, rPET(5HT1a)=.75, rPET(5HT2a)=.81; nearest neighbour with smoothing: rPET(GABA_a)=-.068, rPET(mGluR5)=.05, rPET(5HT1a)=.62, rPET(5HT2a)=.2). To measure spatial changes of gene expression, we derived the gradient of kriged gene expression, i.e. vector field capturing direction and magnitude of local maximal changes. We located sinks and sources of spatial changes via the divergence of the gradient for each gene, which was remarkably similar across all pairs of genes (r(7562 x 7562)=0.87 to 1.0). At every vertex, we assessed the alignment across genes by enveloping gradient vectors with a tensor, then calculated vertex-wise multi-gene fractional anisotropy (range=.69 to 1). Multi-gene anisotropy followed the direction of brain morphology, namely sulcal depth (mode minimal angle=13) and mean curvature (mode minimal angle=14), and significantly (pspins(1000)<.05) correlated with curated measures of structural and functional stability and maturation, namely inter-subject variability of functional topography (r=-.33), allometric scaling (r=-.32 to -.25), evolutionary expansion (r=-.21) myelin content (r=-.18) and cortical thickness (r=-.16).
HBHL Symposium 