Recent advancements in imaging across scales has propelled the development of anatomic brain atlases reflecting a wide range of cellular properties. It’s important to note however that such atlases do not describe function beyond the level of the individual cell. Two additional components are needed to understand functional organization. The first is to understand how chemical information such as neurotransmitters modulate function at both the cellular and circuit level. The second is to understand how densely connected cells, are tuned to different input/outputs and modulate their activity accordingly. These latter two components of brain organization introduce functional flexibility within a fixed infrastructure. At the macroscopic scale, imaging typically depicts how ensembles of neurons work together to execute a task, forming activated or connected regions with millions of neurons within each region. This comment highlights the complexities and challenges associated with understanding function in the context of structure. Inconsistent definitions of function across scales are revealed as a source of confusion when neuroscientists discuss structure-function relationships. There are at least four distinct levels of brain function, ranging from microscopic neural properties to macroscopic functional networks comprised of ensembles of neurons working together. Care must be taken at each scale to understand how structure and function interact. In this comment we emphasize the important role of brain atlases in neuroscience for orientation, comparison, and data reduction. As we develop structural atlases it is essential to understand how flexible function arises within. this fixed infrastructure. We advocate for a nuanced understanding of when and how to integrate structure and function across scales. This work underscores the necessity of reconciling various functional definitions from individual cells to ensemble activity to advance brain science and improve our understanding of structure-function relationships.

While disordered brain circuits should manifest in psychiatric symptoms and cognitive deficits, how they jointly impact multiple behaviors remains poorly understood. Connectome-based predictive (CPM) modeling can identify functional networks associated with specific behavioral measures. The derived networks provide evidence of where an individual’s disordered circuits are, and prediction strength indexes network modeling accuracy. Here, we used CPM to predict a broad range of self-reported clinical and objective cognitive measures in a large, transdiagnostic sample with extensive fMRI data (n = 317). Prediction performance varies substantially across instruments, with cognitive tests yielding stronger models than clinical measures (p < 0.001). To test whether circuits underlying cognitive deficits associated with symptomatology reside in regions where networks overlap, we constructed predictive models using these sparsely shared circuits. They strongly predict cognitive performance and are primarily localized within the frontoparietal network and between the frontoparietal and default mode networks. These findings demonstrate that constraining predictive models to features shared between multiple phenotypes in training data can improve network localization precision. Simon et al. leverage connectome-based predictive modeling to identify brain circuits related to cognitive deficits that co-occur with self-reported psychiatric symptoms. They find that the sparsely shared networks strongly predict cognition and are often localized within the frontoparietal and between the frontoparietal and default mode networks.

We depart from the feed-forward approach of brain-behavior modeling, which identifies the functional brain connectivity networks associated with performance on external tests, and instead introduce a feedback approach that reveals the brain systems those external tests reflect. In fMRI data from n = 302 demographically and clinically diverse participants, we a priori define connectivity networks for six cognitive constructs and employ kernel ridge regression to quantify each network's contribution to test performance. This approach provides a ranking of test scores according to the predictive power of each cognitive network, revealing which tests probe which brain networks. It further identifies combinations of measures that optimally probe predefined brain systems and evaluates how specific subtests influence composite scores, revealing when subset inclusion reinforces or weakens specific brain circuit and composite score relationships. This work opens an avenue of research by providing a framework for the development of test instruments guided by quantitative brain metrics.

Excitation using nonlinear gradient magnetic fields is investigated as a means of sub-volume magnetic resonance imaging (MRI). Conventional gradient fields provide encoding along a single direction, whereas nonlinear gradient fields encode information simultaneously along at least two directions. This leads to excitation regions (FOX) that have curvilinear boundaries, which may be more tolerant to aliasing artifacts when the encoded field of view (FOV) is smaller than the FOX. This reduces the complexity of the required radiofrequency (RF) excitation pulses and enables accelerated reduced-FOV imaging with standard slice-selection RF-pulses. We demonstrate the approach using a Z2-harmonic field for cylindrical regions of interest (ROIs) with various radius/height ratios. The minimum-FOV that should be encoded is formulated in terms of ROI and RF pulse parameters to allow a theoretical evaluation of feasibility during study design. The investigated method is compared to one-dimensional and two-dimensional selective RF pulses in terms of echo time, scan time and specific absorption rate (SAR) using simulations and phantom experiments. The investigated method yields lower scan time while keeping the SAR unaltered compared to a conventional slice-selective RF pulse, and is more efficient in terms of SAR, echo time and scan time compared to two-dimensional selective excitation.

Autism is a heterogeneous condition, and functional magnetic resonance imaging-based studies have advanced understanding of neurobiological correlates of autistic features. Little work has focused on the optimal brain states to reveal brain-phenotype relationships. Here, using connectome-based predictive modeling, we interrogated four datasets to determine scanning conditions that boost prediction of clinically relevant phenotypes and assess generalizability. In dataset one, a sample of youth with autism and neurotypical participants (n = 63), we found that a sustained attention task resulted in high prediction performance of autistic traits compared with a free-viewing social attention task and a resting-state condition. In dataset two (n = 25), we observed the predictive network model of autistic traits generated from the sustained attention task generalized to predict measures of attention in neurotypical adults. In datasets three and four, we determined the same predictive network model further generalized to predict measures of social responsiveness in the Autism Brain Imaging Data Exchange (n = 229) and the Healthy Brain Network (n = 643). Our data suggest an in-scanner sustained attention challenge can help delineate robust markers of autistic traits.

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) plays an important role in breast cancer screening, tumor assessment, and treatment planning and monitoring. The dynamic changes in contrast in different tissues help to highlight the tumor in post-contrast images. However, varying acquisition protocols and individual factors result in large variation in the appearance of tissues, even for images acquired in the same phase (e.g., first post-contrast phase), making automated tumor segmentation challenging. Here, we propose a tumor segmentation method that leverages knowledge of the image acquisition time to modulate model features according to the specific acquisition sequence. We incorporate the acquisition times using feature-wise linear modulation (FiLM) layers, a lightweight method for incorporating temporal information that also allows for capitalizing on the full, variables number of images acquired per imaging study. We trained baseline and different configurations for the time-modulated models with varying backbone architectures on a large public multisite breast DCE-MRI dataset. Evaluation on in-domain images and a public out-of-domain dataset showed that incorporating knowledge of phase acquisition time improved tumor segmentation performance and model generalization.

Low-field (LF) magnetic resonance imaging (MRI) improves accessibility and reduces costs but generally has lower signal-to-noise ratios and degraded contrast compared to high field (HF) MRI, limiting its clinical utility. Simulating LF MRI from HF MRI enables virtual evaluation of novel imaging devices and development of LF algorithms. Existing low field simulators rely on noise injection and smoothing, which fail to capture the contrast degradation seen in LF acquisitions. To this end, we introduce an end-to-end LF-MRI synthesis framework that learns HF to LF image degradation directly from a small number of paired HF-LF MRIs. Specifically, we introduce a novel HF to LF coordinate-image decoupled neural operator (H2LO) to model the underlying degradation process, and tailor it to capture high-frequency noise textures and image structure. Experimental results in T1w and T2w MRI demonstrate that H2LO produces more faithful simulated low-field images than existing parameterized noise synthesis models and popular image-to-image translation models. Furthermore, it improves performance in downstream image enhancement tasks, showcasing its potential to enhance LF MRI diagnostic capabilities.

Neuroscience aims to understand how the structural and functional organization of the brain relates to behavior. Structural studies at the cellular level establish the neurobiological framework for understanding function. Structure is only part of the story however, with molecular and functional information needed for a comprehensive view. At the mesoscopic level, functional regions are defined by ensembles of neurons that work together. This perspective reviews the relationship between structural and functional organization across scales and aims to provide a common language for neuroscientists working at any level. While it is unequivocal that structure constrains function at the cellular level, understanding function at the meso- and macro-scopic scales is much more complicated involving ensembles of neurons and their dynamic interactions. Recognition of these scale differences is essential for advancing representational models in the field.

PURPOSE:To evaluate k-space acquisition strategies for magnetic resonance imaging (MRI) in a nonuniform B0 (NuBo) field-cycling system, focusing on image quality, scan-time efficiency, robustness to B0 inhomogeneity, and RF coil bandwidth constraints. METHODS:Three acquisition strategies were compared: single-shot turbo spin echo (SS-TSE), multi-echo spin echo (MESE) with repeated phase encoding, and spin echo single-point imaging (SE-SPI) without applied readout (RO) gradients. Experiments were performed on an open, low-field, field-cycling MRI system using a nonuniform electromagnet. Imaging performance was assessed using a phantom and biological samples under matched total scan-time conditions. RF coils with wide- and narrow-bandwidth designs were evaluated to study sensitivity trade-offs. Retrospective k-space undersampling and compressed sensing reconstruction were applied to investigate potential scan-time reduction for all acquisition strategies. RESULTS:SE-SPI consistently achieved larger full width at half maximum (FWHM) values than SS-TSE and MESE, indicating improved edge definition. Edge sharpness was higher along the RO (PE1) direction but comparable along the PE (PE2) direction. SS-TSE provided the fastest k-space acquisition and the highest central k-space SNR but exhibited increased blurring and bandwidth-related signal loss. MESE improved edge definition over SS-TSE while requiring additional polarization cycles. SE-SPI also showed greater robustness to compressed sensing undersampling. CONCLUSION:SE-SPI provides a robust acquisition strategy for MRI in nonuniform B0 field-cycling systems by improving image fidelity, maintaining uniform signal intensity, enabling efficient use of high-Q RF coils, and supporting compressed sensing acceleration.

Sustained attention is an important neurobiological process. Difficulties with attention play a key role in neurodevelopmental disorders, such as attention-deficit/hyperactivity disorder (ADHD) and autism. Here, we identified functional connections consistently associated with sustained attention across datasets, participant populations, and fMRI scan types. We interrogated five transdiagnostic, previously published connectome-based models predicting attention and autistic phenotypes. All models were related to sustained attention, including in samples comprising participants with autism. As expected, we observed that models predicting attention phenotypes shared more similar features with each other than models predicting autism symptoms. Interestingly, we observed no statistically significant model similarities when considering factors such as age, functional run type, or diagnosis. This suggests that functional connectivity patterns predicting individual differences in behavior tend to be phenotype-specific, regardless of age or clinical diagnosis. Our results underscore the importance of searching for consistent markers of transdiagnostic sustained attention phenotypes in neurodevelopmental conditions.

Cognitive control supports adaptive responses in an ever-changing world. While alterations in cognitive control have been consistently observed in a range of psychiatric disorders, the neural mechanisms giving rise to this behavioral variation remain elusive. Here, we tested whether the ability to flexibly recruit recurring brain activation patterns (i.e., brain states) may serve as an intermediate phenotype supporting cognitive control in individuals with a spectrum of clinical symptoms. We leveraged machine learning and external validation to explore this question in three independent, transdiagnostic datasets (N>600), including participants with anxiety disorders, schizophrenia, mood disorders, substance use disorders, post-traumatic stress disorder, obsessive-compulsive disorder, and neurodevelopmental disorders. To capture cognitive control’s multifaceted nature, we assessed two of its components—inhibition and shift— using both task-based and questionnaire data. Flexible brain state engagement predicted all cognitive control metrics in previously unseen individuals transdiagnostically, regardless of which dataset was used for model training. Connectome-based predictive modeling also revealed that shared brain networks underpinned flexible brain state engagement in a transdiagnostic manner. Leveraging brain network dynamics, we further observed that moments of more flexible brain state engagement aligned with moments of network connectivity related to better cognitive control within the same individual. This temporal alignment was replicated in all three datasets with heterogeneous samples. Altogether, this study suggests flexible engagement of brain states may support both inter- and intra-individual differences in cognitive control across individuals with diverse mental health profiles.

Popular methods for analyzing the brain's functional connectome examine statistical associations between pairs of atlas-defined brain regions, viewing the strength of these links as independent values. However, edges within a standard connectivity matrix, that is, correlations between individual regions or nodes, are not independent. They are part of an interconnected system. Here, we propose that consideration of both independent, linear relationships (as in standard approaches such as linear kernel ridge regression and connectome-based predictive modeling) and higher order statistical associations-such as tertiary interactions between matrix components and global features of the matrix space-will enhance identification of meaningful individual differences. To test this, we adopt a geometrically grounded measure of similarity that accounts for higher-order local statistical relationships and global interactions, the Wasserstein metric. Results indicate that considering connectivity matrices as representations of their associated Gaussian distributions significantly improves identification of individuals based on their connectivity matrices (aka, "fingerprinting"). We further show that when incorporated into our novel pipeline, "connectome-regression in any metric (CRAM)" the Wasserstein and (the CRAM pipeline itself) improve prediction of individual differences in phenotypes such as fluid intelligence and openness to experience. Thus, both pairwise local and global brain connectivity properties encode for meaningful individual differences that relate to phenotypic expressions and should be considered in brain-behavior predictive models.

Each individual's complex, multidimensional environment, known as their "exposome", plays an essential role in shaping cognitive neurodevelopment. Understanding the mechanisms whereby children's exposome influences their development is crucial to facilitate the design of interventions to foster positive developmental trajectories for all youth. Recent work has identified a general exposome factor associated with socio-economic inequality that is strongly related to cognition and individual differences in the spatial organization of functional brain networks in youth. Building on these findings, the current study explores whether alterations in functional connectivity may represent a potential mechanism linking variation in the exposome to cognitive performance. We apply a data-driven, cross-validated, whole-brain machine learning approach, connectome-based statistical inference, to identify patterns of functional connectivity associated with exposome scores among early adolescents enrolled in the Adolescent Brain Cognitive Development (ABCD) Study using data collected during three cognitive tasks and during rest. Additionally, we investigate whether the identified patterns of functional connectivity relate to individual differences in cognitive performance across three domains: General Cognition, Executive Functioning, and Learning/Memory. Models incorporating 10-fold cross-validation over 100 iterations identified consistent functional connections associated with the exposome across task and rest conditions (model performance: ns = 6137 - 8391, rs = 0.34-0.44, ps < .001). Results were robust across data collection sites and functional connections common across all significant models were associated with cognitive performance across domains (ps < 0.0009). Collectively, these findings reveal that multidimensional environmental exposures are reflected in patterns of functional connectivity and relate to cognitive functioning among youth.

Modern neuroimaging techniques have enabled great strides in understanding how the brain gives rise to behavior. While uncovering relationships between neural systems and psychiatric illnesses offers great promise for improving patient outcomes, current therapeutic approaches that utilize brain-behavior associations remain largely experimental and have seen limited integration into routine clinical practice. In this review, we outline key challenges and opportunities across 3 main stages in the brain-behavior therapeutic development cycle: identification, interpretation, and implementation. First, we discuss several experimental and technical considerations when elucidating associations between the brain (regions, networks, dynamics, etc.) and clinically relevant behaviors, with an emphasis on multimodal imaging and novel psychometric approaches. Next, we explore the complexities of interpreting brain-behavior relationships to glean mechanistic insights for therapeutic development. These include modeling considerations, causal inference, and contending with weak associations. Finally, we discuss the barriers facing clinical implementation spanning technical, political, and ethical dimensions. By laying out what we consider to be the critical challenges impeding therapeutic development from start to finish, alongside the potential opportunities to address them, we hope to accelerate the progression of brain-behavior associations through identification, interpretation, and implementation, culminating in new impactful methods for treating psychiatric illnesses.

Cocaine use disorder (CUD) remains a major public health problem, particularly among individuals receiving medications for opioid use disorder (mOUD), where ongoing cocaine use is associated with poorer treatment outcomes and elevated overdose risk. N-acetylcysteine (NAC) is a promising adjunctive candidate that reduces cocaine-seeking in preclinical models and shows mixed efficacy in humans, but its functional neural mechanisms in individuals with CUD—especially within mOUD settings—remain poorly understood. The present study uses functional magnetic resonance imaging (fMRI) to determine whether short-term NAC administration modulates brain activity in regions associated with inhibitory control and affective processing in people with CUD receiving medications for opioid use disorder (mOUD). In this double-blind, placebo-controlled, crossover study, 21 individuals with CUD enrolled in an mOUD program were randomized to receive NAC or placebo for 7 days and participated in fMRI scanning on the 7th day of this period. Following a fourteen-day washout period, participants were crossed over to receive the other condition for another 7-day period. Participants received a second fMRI scan on the 7th day of this period. Participants completed a go/no-go (GNG) and emotion-regulation task (ERT) during each scan. Compared to placebo, active NAC did not significantly affect brain activity during either the GNG task or the ERT. Overall, short-term NAC did not robustly modulate neural circuits supporting inhibitory control or emotion regulation in individuals with comorbid CUD and OUD receiving mOUD, consistent with prior trials reporting limited efficacy of NAC for CUD. These findings underscore the value of incorporating neuroimaging into early-phase treatment trials to evaluate target engagement in humans. ClinicalTrials.gov identifier: NCT02994875, registered 2016–12-16.

YaleNeuroConnect is a human functional MRI (fMRI) dataset collected at Yale University that includes functional MRI data (and the respective functional connectomes) obtained under resting-state and six task conditions. There are 302 diagnostically and demographically diverse subjects, each with extensive neuropsychological testing and symptom inventories obtained outside of the MRI. Prior studies have shown that stronger predictive models relating the brain to external measures can be built with connectivity data obtained during continuous performance tasks instead of the more common resting-state. The tasks here were selected to exercise the brain across various cognitive domains. For each subject, 48 minutes of fMRI data and high-resolution 3D brain volumes were obtained. The fMRI data, along with the deep phenotyping data in a diverse subject pool, allow studies of brain parcellation under different conditions, the relationship between cognitive and clinical measures, identification of circuits supporting external measures, and data for the development of brain-based tests. The transdiagnostic nature of the sample allows a sufficient range of symptom scores to test the principles of the Research Domain Criteria framework.

The mammalian brain is comprised of anatomically and functionally distinct regions. Substantial work over the past century has pursued the generation of ever-more accurate maps of regional boundaries, using either expert judgement or data-driven clustering of functional, connectional, and/or architectonic properties. However, these approaches are often purely descriptive, have limited generalizability, and do not elucidate the underlying generative mechanisms that shape the regional organization of the brain. Here, we develop a novel approach that leverages a simple, hierarchical principle for generating a multiscale parcellation of any brain structure in any mammalian species using only its geometry. We show that this approach yields regions at any resolution scale that are more homogeneous than those defined in nearly all existing benchmark brain parcellations in use today across hundreds of anatomical, functional, cellular, and molecular brain properties measured in humans, macaques, marmosets, and mice. We additionally show how our method can be generalized to previously unstudied mammalian species for which no parcellations exist. Finally, we demonstrate how our approach captures the essence of a simple, hierarchical reaction-diffusion mechanism, in which the geometry of a brain structure shapes the spatial expression of putative patterning molecules linked to the formation of distinct regions through development. Our findings point to a highly conserved and universal influence of geometry on the regional organization of the mammalian brain.