Mitigating site effects in covariance for machine learning in neuroimaging data

Comprehensive evaluation of pipelines for classification of psychiatric disorders using multi-site resting-state fMRI datasets

Objective classification biomarkers that are developed using resting-state functional magnetic resonance imaging (rs-fMRI) data are expected to contribute to more effective treatment for psychiatric disorders. Unfortunately, no widely accepted biomarkers are available at present, partially because of the large variety of analysis pipelines for their development. In this study, we comprehensively evaluated analysis pipelines using a large-scale, multi-site fMRI dataset for major depressive disorder (MDD). We explored combinations of options in four sub-processes of the analysis pipelines: six types of brain parcellation, four types of functional connectivity (FC) estimations, three types of site-difference harmonization, and five types of machine-learning methods. A total of 360 different MDD classification biomarkers were constructed using the SRPBS dataset acquired with unified protocols (713 participants from four sites) as the discovery dataset, and datasets from other projects acquired with heterogeneous protocols (449 participants from four sites) were used for independent validation. We repeated the procedure after swapping the roles of the two datasets to identify superior pipelines, regardless of the discovery dataset. The classification results of the top 10 biomarkers showed high similarity, and weight similarity was observed between eight of the biomarkers, except for two that used both data-driven parcellation and FC computation. We applied the top 10 pipelines to the datasets of other psychiatric disorders (autism spectrum disorder and schizophrenia), and eight of the biomarkers exhibited sufficient classification performance for both disorders. Our results will be useful for establishing a standardized pipeline for classification biomarkers.

Rethinking MRI as a measurement device through modular and portable pipelines

The premise of MRI as a reliable measurement device is limited by proprietary barriers and inconsistent implementations, which prevent the establishment of measurement uncertainties. As a result, biomedical studies that rely on these methods are plagued by systematic variance, undermining the perceived promise of quantitative imaging biomarkers (QIBs) and hindering their clinical translation. This review explores the added value of open-source measurement pipelines in minimizing variability sources that would otherwise remain unknown. First, we introduce a tiered benchmarking framework (from black-box to glass-box) that exposes how opacity at different workflow stages propagates measurement uncertainty. Second, we provide a concise glossary to promote consistent terminology for strategies that enhance reproducibility before acquisition or enable valid post-hoc pooling of QIBs. Building on this foundation, we present two illustrative measurement workflows that decouple workflow logic from the orchestration of computational processes in an MRI measurement pipeline, rooted in the core principles of modularity and portability. Designed as accessible entry points for implementation, these examples serve as practical guides, helping users adapt the frameworks to their specific needs and facilitating collaboration. Through critical evaluation of existing approaches, we discuss how standardized workflows can help identify outstanding challenges in translating glass-box frameworks into clinical scanner environments. Ultimately, achieving this goal will require coordinated efforts from QIB developers, regulators, industry partners, and clinicians alike.

A multi-site, multi-modal travelling-heads resource for brain MRI harmonisation

Despite its great potential for studying the living brain, magnetic resonance imaging (MRI) can be often limited by nuisance non-biological factors, such as hardware/software differences between scanners, which can interfere with biological variability. This lack of standardisation or harmonisation between scanners hinders reproducibility and quantifiability of MRI. Towards addressing this challenge, we present one of the most comprehensive MRI harmonisation resources, based on a travelling heads paradigm; healthy volunteers scanned repeatedly across different scanners. The Oxford-Nottingham Harmonisation (ON-Harmony) resource offers data from 20 participants each scanned on six different 3 T MRI scanners from three major vendors (GE/Philips/Siemens) across five imaging sites. Each scanning session includes five imaging modalities (T1w/T2w/dMRI/rfMRI/SWI) with protocols aligned to the UK Biobank, while for about half of the participants five within-scanner repeats are additionally acquired. The 165 multi-modal scanning sessions allow mapping of different pools of variability (biological, between-scanner, within-scanner) for hundreds of MRI-derived measures. We describe the breadth of information contained in the publicly-available data and showcase their reuse potential for evaluating efficacy of harmonisation approaches.

Two Axes of White Matter Development

Despite decades of neuroimaging research, how white matter develops along the length of major tracts in humans remains unknown. Here, we identify fundamental patterns of white matter maturation by examining developmental variation along major, long-range cortico-cortical tracts in youth ages 5-23 years using diffusion MRI from three large-scale, cross-sectional datasets (total N = 2,710). Across datasets, we delineate two replicable axes of human white matter development. First, we find a deep-to-superficial axis, in which superficial tract regions near the cortical surface exhibit greater age-related change than deep tract regions. Second, we demonstrate that the development of superficial tract regions aligns with the cortical hierarchy defined by the sensorimotor-association axis, with tract ends adjacent to sensorimotor cortices maturing earlier than those adjacent to association cortices. These results reveal developmental variation along tracts that conventional tract-average analyses have previously obscured, challenging the implicit assumption that white matter tracts mature uniformly along their length. Such developmental variation along tracts may have functional implications, including mitigating ephaptic coupling in densely packed deep tract regions and tuning neural synchrony through hierarchical development in superficial tract regions - ultimately refining neural transmission in youth.

The impact of the novel CovBat harmonization method on enhancing radiomics feature stability and machine learning model performance: A multi-center, multi-device study

PURPOSE

This study aims to assess whether the novel CovBat harmonization method can further reduce radiomics feature variability from different imaging devices in multi-center studies and improve machine learning model performance compared to the ComBat method.

MATERIALS

Non-contrast abdominal CT scans of 1,000 healthy subjects from three medical institutions (from four manufacturers and eight different models) were retrospectively included: Hospital A (n = 513), Hospital B (n = 338), and Hospital C (n = 149). 93 radiomics features were extracted from liver and spleen tissues using PyRadiomics. Performing a binary classification task of liver and spleen tissues on the pooled data from the three institutions: (1) Unharmonized, (2) ComBat, and (3) CovBat. Models were built separately for each radiomics feature classes (First-order, GLCM, GLRLM, GLSZM, NGTD, GLDM), as well as a combined model integrating all feature classes. The Kruskal-Wallis test and principal component analysis (PCA) were used to assess the variability of radiomics features among the groups. Multiple linear regression models were used to analyze the sources of variation. Accuracy, sensitivity, specificity, F1-score, and area under the curve (AUC) were used to evaluate model performance.

RESULTS

After ComBat and CovBat harmonization, the number of consistent features increased by 68.82 % and 73.12 %, respectively, and the feature variability due to hardware differences decreased from 12.32-25.38 % to 1.89-2.01 % with ComBat and 1.19-1.88 % with CovBat. The AUC of the machine learning models improved significantly: Combined (Unharmonized: 0.93, ComBat: 0.99, CovBat: 1.00), First-order (0.93, 0.98, 0.98), GLCM (0.81, 0.93, 0.98), GLRLM (0.78, 0.96, 0.98), NGTDM (0.75, 0.96, 0.98), GLSZM (0.78, 0.93, 0.97), and GLDM (0.83, 0.94, 0.97). DeLong's test showed that the results before and after harmonization were statistically significant (P < 0.05).

CONCLUSION

CovBat further reduced radiomics feature variability caused by different CT scanners and significantly improved the performance of machine learning models, although the degree of improvement varied across different feature categories.

Reproducible Brain Charts: An open data resource for mapping brain development and its associations with mental health

Major mental disorders are increasingly understood as disorders of brain development. Large and heterogeneous samples are required to define generalizable links between brain development and psychopathology. To this end, we introduce the Reproducible Brain Charts (RBC), an open data resource that integrates data from 5 large studies of brain development in youth from three continents (N=6,346; 45% Female). Confirmatory bifactor models were used to create harmonized psychiatric phenotypes that capture major dimensions of psychopathology. Following rigorous quality assurance, neuroimaging data were carefully curated and processed using consistent pipelines in a reproducible manner with DataLad, the Configurable Pipeline for the Analysis of Connectomes (C-PAC), and FreeSurfer. Initial analyses of RBC data emphasize the benefit of careful quality assurance and data harmonization in delineating developmental effects and associations with psychopathology. Critically, all RBC data - including harmonized psychiatric phenotypes, unprocessed images, and fully processed imaging derivatives - are openly shared without a data use agreement via the International Neuroimaging Data-sharing Initiative. Together, RBC facilitates large-scale, reproducible, and generalizable research in developmental and psychiatric neuroscience.

Addressing Inter-reconstruction variability in multi-energy myocardial CT Radiomics: The Benefits of combat harmonization

RATIONALE AND OBJECTIVES

To investigate the effect of ComBat harmonization on the stability of myocardial radiomic features derived from multi-energy CT reconstructions.

MATERIALS AND METHODS

A retrospective study was conducted on 205 patients who underwent dual-energy chest CTA at a single center. The data was reconstructed into multiple spectral reconstructions (mixed energy simulating standard 120 Kv acquisition and monoenergetic images ranging from 40 to 190 keV in increments of 10). The left ventricle myocardium was segmented using semiautomated software (Syngo.Via FRONTIER, version 5.0.2; Siemens). Radiomic features were extracted from multiple spectral reconstructions (batches). The consistency of these radiomics features across different batches was evaluated with and without ComBat harmonization using Cohen's d and Principal component analysis (PCA). Both parametric and nonparametric ComBat methods were considered.

RESULTS

Without any ComBat technique, 43.40% of features remained consistent across all multienergy reconstructions. Applying ComBat harmonization increased this consistency to 98.37% with parametric empirical bayes (EB) ComBat and EB M-ComBat, and to 91.52% and 92.33% with nonparametric EB ComBat and nonparametric EB M-ComBat, respectively. PCA without ComBat revealed noticeable differences in the first two principal components between batches, indicating a batch effect or unstable radiomic features. Following ComBat harmonization, the principal components showed more consistency between batches, demonstrating radiomics feature stability between batches.

CONCLUSION

ComBat harmonization enhanced the consistency of radiomic features from multi-energy CT data. Integrating ComBat harmonization may lead to more reproducible results in multienergy CT radiomics studies.

Increased structural covariance of cortical measures in individuals with an at-risk mental state

An anomalous pattern of structural covariance has been reported in schizophrenia, which has been suggested to represent connectome changes during brain maturation and neuroprogressive processes. It remains unclear whether similar differences exist in a clinical high-risk state for psychosis, and if they are associated with a prodromal phenotype and/or later psychosis onset. This multicenter magnetic resonance imaging study cross-sectionally examined structural covariance in a large at-risk mental state (ARMS) sample with different outcomes. The whole-brain structural covariance of four cortical measures (thickness, area, volume, and gyrification) was assessed in 155 individuals with ARMS, who were subclassified into 26 (16.8 %) with a later psychosis onset (ARMS-P), 44 with persistent subthreshold psychotic symptoms, and 53 with the remission of psychotic symptoms (ARMS-R) during the clinical follow-up, and 191 healthy controls. The relationships of changes in structural covariance with clinical symptoms and cognitive impairments were also investigated in the ARMS subsample. Structural covariance was significantly higher in widespread cortical regions in the ARMS group than in the controls, with each cortical measure having a different pattern in affected cortical regions. The higher structural covariance of the cortical area was partly related to severe suspiciousness-persecutory ideation. Structural covariance was significantly higher, mainly in fronto-parietal gyrification, in the ARMS-P group than in the ARMS-R group. The present results suggest that changes in structural covariance result in psychosis vulnerability and the excessive structural covariance of brain gyrification in ARMS subjects may contribute to their later clinical course.

DeepResBat: Deep residual batch harmonization accounting for covariate distribution differences

Pooling MRI data from multiple datasets requires harmonization to reduce undesired inter-site variabilities, while preserving effects of biological variables (or covariates). The popular harmonization approach ComBat uses a mixed effect regression framework that explicitly accounts for covariate distribution differences across datasets. There is also significant interest in developing harmonization approaches based on deep neural networks (DNNs), such as conditional variational autoencoder (cVAE). However, current DNN approaches do not explicitly account for covariate distribution differences across datasets. Here, we provide mathematical results, suggesting that not accounting for covariates can lead to suboptimal harmonization. We propose two DNN-based covariate-aware harmonization approaches: covariate VAE (coVAE) and DeepResBat. The coVAE approach is a natural extension of cVAE by concatenating covariates and site information with site- and covariate-invariant latent representations. DeepResBat adopts a residual framework inspired by ComBat. DeepResBat first removes the effects of covariates with nonlinear regression trees, followed by eliminating site differences with cVAE. Finally, covariate effects are added back to the harmonized residuals. Using three datasets from three continents with a total of 2787 participants and 10,085 anatomical T1 scans, we find that DeepResBat and coVAE outperformed ComBat, CovBat and cVAE in terms of removing dataset differences, while enhancing biological effects of interest. However, coVAE hallucinates spurious associations between anatomical MRI and covariates even when no association exists. Future studies proposing DNN-based harmonization approaches should be aware of this false positive pitfall. Overall, our results suggest that DeepResBat is an effective deep learning alternative to ComBat. Code for DeepResBat can be found here: https://github.com/ThomasYeoLab/CBIG/tree/master/stable_projects/harmonization/An2024_DeepResBat.

Healthy core: Harmonizing brain MRI for supporting multicenter migraine classification studies

Multicenter and multi-scanner imaging studies may be necessary to ensure sufficiently large sample sizes for developing accurate predictive models. However, multicenter studies, incorporating varying research participant characteristics, MRI scanners, and imaging acquisition protocols, may introduce confounding factors, potentially hindering the creation of generalizable machine learning models. Models developed using one dataset may not readily apply to another, emphasizing the importance of classification model generalizability in multi-scanner and multicenter studies for producing reproducible results. This study focuses on enhancing generalizability in classifying individual migraine patients and healthy controls using brain MRI data through a data harmonization strategy. We propose identifying a 'healthy core'-a group of homogeneous healthy controls with similar characteristics-from multicenter studies. The Maximum Mean Discrepancy (MMD) in Geodesic Flow Kernel (GFK) space is employed to compare two datasets, capturing data variabilities and facilitating the identification of this 'healthy core'. Homogeneous healthy controls play a vital role in mitigating unwanted heterogeneity, enabling the development of highly accurate classification models with improved performance on new datasets. Extensive experimental results underscore the benefits of leveraging a 'healthy core'. We utilized two datasets: one comprising 120 individuals (66 with migraine and 54 healthy controls), and another comprising 76 individuals (34 with migraine and 42 healthy controls). Notably, a homogeneous dataset derived from a cohort of healthy controls yielded a significant 25% accuracy improvement for both episodic and chronic migraineurs.

Assessment of ComBat Harmonization Performance on Structural Magnetic Resonance Imaging Measurements

Data aggregation across multiple research centers is gaining importance in the context of MRI research, driving diverse high-dimensional datasets to form large-scale heterogeneous sample, increasing statistical power and relevance of machine learning and deep learning algorithm. Site-related effects have been demonstrated to introduce bias in MRI features and confound subsequent analyses. Although Combating Batch (ComBat) technique has been recently reported to successfully harmonize multi-scale neuroimaging features, its performance assessments are still limited and largely based on qualitative visualizations and statistical analyses. In this study, we stand out by using a robust cross-validation approach to assess ComBat performances applied on volume- and surface-based measures acquired across three sites. A machine learning approach based on Multi-Class Gaussian Process Classifier was applied to predict imaging site based on raw and harmonized brain features, providing quantitative insights into ComBat effectiveness, and verifying the association between biological covariates and harmonized brain features. Our findings showed differences in terms of ComBat performances across measures of regional brain morphology, demonstrating tissue specific site effect modeling. ComBat adjustment of site effects also varied across regional level of each specific volume-based and surface-based measures. ComBat effectively eliminates unwanted data site-related variability, by maintaining or even enhancing data association with biological factors. Of note, ComBat has demonstrated flexibility and robustness of application on unseen independent gray matter volume data from the same sites.

Increasing hub disruption parallels dementia severity in autosomal dominant Alzheimer's disease

Hub regions in the brain, recognized for their roles in ensuring efficient information transfer, are vulnerable to pathological alterations in neurodegenerative conditions, including Alzheimer's disease (AD). Computational simulations and animal experiments have hinted at the theory of activity-dependent degeneration as the cause of this hub vulnerability. However, two critical issues remain unresolved. First, past research has not clearly distinguished between two scenarios: hub regions facing a higher risk of connectivity disruption (targeted attack) and all regions having an equal risk (random attack). Second, human studies offering support for activity-dependent explanations remain scarce. We refined the hub disruption index to demonstrate a hub disruption pattern in functional connectivity in autosomal dominant AD that aligned with targeted attacks. This hub disruption is detectable even in preclinical stages, 12 years before the expected symptom onset and is amplified alongside symptomatic progression. Moreover, hub disruption was primarily tied to regional differences in global connectivity and sequentially followed changes observed in amyloid-beta positron emission tomography cortical markers, consistent with the activity-dependent degeneration explanation. Taken together, our findings deepen the understanding of brain network organization in neurodegenerative diseases and could be instrumental in refining diagnostic and targeted therapeutic strategies for AD in the future.

A data integration method for new advances in development cognitive neuroscience

Combining existing datasets to investigate key questions in developmental cognitive neuroscience brings exciting opportunities and unique challenges. However, many data pooling methods require identical or harmonized methodologies that are often not feasible. We propose Integrative Data Analysis (IDA) as a promising framework to advance developmental cognitive neuroscience with secondary data analysis. IDA serves to test hypotheses by combining data of the same construct from commensurate (but not identical) measures. To overcome idiosyncrasies of neuroimaging data, IDA explicitly evaluates if measures across studies assess the same construct. Moreover, IDA allows investigators to examine meaningful individual variability by de-confounding source-specific differences. To demonstrate IDA's potential, we explain foundational concepts, outline necessary steps, and apply IDA to volumetric measures of hippocampal subfields from 443 4- to 17-year-olds across three independent studies. We identified commensurate measures of Cornu Ammonis (CA) 1, dentate gyrus (DG)/CA3, and Subiculum (Sub). Model testing supported use of IDA to create IDA factor scores. We found age-related differences in DG/CA3, not but CA1 and Sub volume in the integrated dataset. By successfully demonstrating IDA, our hope is that future innovations come from the combination of existing neuroimaging data to create representative integrated samples when testing critical developmental questions.

Harmonization for Parkinson's Disease Multi-Dataset T1 MRI Morphometry Classification

Classification of disease and healthy volunteer cohorts provides a useful clinical alternative to traditional group statistics due to individualized, personalized predictions. Classifiers for neurodegenerative disease can be trained on structural MRI morphometry, but require large multi-scanner datasets, introducing confounding batch effects. We test ComBat, a common harmonization model, in an example application to classify subjects with Parkinson's disease from healthy volunteers and identify common pitfalls, including data leakage. We used a multi-dataset cohort of 372 subjects (216 with Parkinson's disease, 156 healthy volunteers) from 11 identified scanners. We extracted both FreeSurfer and the determinant of Jacobian morphometry to compare single-scanner and multi-scanner classification pipelines. We confirm the presence of batch effects by running single scanner classifiers which could achieve wildly divergent AUCs on scanner-specific datasets (mean:0.651 ± 0.144). Multi-scanner classifiers that considered neurobiological batch effects between sites could easily achieve a test AUC of 0.902, though pipelines that prevented data leakage could only achieve a test AUC of 0.550. We conclude that batch effects remain a major issue for classification problems, such that even impressive single-scanner classifiers are unlikely to generalize to multiple scanners, and that solving for batch effects in a classifier problem must avoid circularity and reporting overly optimistic results.

Detecting and Mitigating the Clever Hans Effect in Medical Imaging: A Scoping Review

The Clever Hans effect occurs when machine learning models rely on spurious correlations instead of clinically relevant features and poses significant challenges to the development of reliable artificial intelligence (AI) systems in medical imaging. This scoping review provides an overview of methods for identifying and addressing the Clever Hans effect in medical imaging AI algorithms. A total of 173 papers published between 2010 and 2024 were reviewed, and 37 articles were selected for detailed analysis, with classification into two categories: detection and mitigation approaches. Detection methods include model-centric, data-centric, and uncertainty and bias-based approaches, while mitigation strategies encompass data manipulation techniques, feature disentanglement and suppression, and domain knowledge-driven approaches. Despite the progress in detecting and mitigating the Clever Hans effect, the majority of current machine learning studies in medical imaging do not report or test for shortcut learning, highlighting the need for more rigorous validation and transparency in AI research. Future research should focus on creating standardized benchmarks, developing automated detection tools, and exploring the integration of detection and mitigation strategies to comprehensively address shortcut learning. Establishing community-driven best practices and leveraging interdisciplinary collaboration will be crucial for ensuring more reliable, generalizable, and equitable AI systems in healthcare.

Revisiting Resting-State Functional Connectivity of the Amygdala and Subgenual Anterior Cingulate Cortex in Adolescents and Adults With Depression

BACKGROUND

Adolescent depression is a growing public health concern, and neuroimaging offers a promising approach to its pathology. We focused on the functional connectivity of the amygdala and subgenual anterior cingulate cortex (sgACC), which is theoretically important in major depressive disorder (MDD), but empirical evidence has remained inconsistent. This discrepancy is likely due to the limited statistical power of small sample sizes.

METHODS

We rigorously examined sgACC-amygdala connectivity in adolescents and adults with depression using data from the Healthy Brain Network (n = 321; 170 female), the ABCD (Adolescent Brain Cognitive Development) Study (n = 141; 56 female), the Boston Adolescent Neuroimaging of Depression and Anxiety study (n = 108; 75 female), and the REST-meta-MDD project (n = 1436; 880 female). Linear mixed models, Bayesian factor analyses, and meta-analysis were used to assess connectivity.

RESULTS

Our analyses revealed that sgACC-amygdala connectivity in adolescents with MDD was comparable to that in healthy control individuals, whereas adults with recurrent MDD exhibited reduced connectivity. Resampling analysis demonstrated that small sample sizes (i.e., n < 30 MDD cases) tend to inflate effects, potentially leading to misinterpretations.

CONCLUSIONS

These findings clarify the state of sgACC-amygdala connectivity in MDD and underscore the importance of refining neurocognitive models separately for adolescents and adults. The study also highlights the necessity for large-scale replication studies to ensure robust and reliable findings.

Relationship between MRI brain-age heterogeneity, cognition, genetics and Alzheimer's disease neuropathology

BACKGROUND

Brain ageing is highly heterogeneous, as it is driven by a variety of normal and neuropathological processes. These processes may differentially affect structural and functional brain ageing across individuals, with more pronounced ageing (older brain age) during midlife being indicative of later development of dementia. Here, we examined whether brain-ageing heterogeneity in unimpaired older adults related to neurodegeneration, different cognitive trajectories, genetic and amyloid-beta (Aβ) profiles, and to predicted progression to Alzheimer's disease (AD).

METHODS

Functional and structural brain age measures were obtained for resting-state functional MRI and structural MRI, respectively, in 3460 cognitively normal individuals across an age range spanning 42-85 years. Participants were categorised into four groups based on the difference between their chronological and predicted age in each modality: advanced age in both (n = 291), resilient in both (n = 260) or advanced in one/resilient in the other (n = 163/153). With the resilient group as the reference, brain-age groups were compared across neuroimaging features of neuropathology (white matter hyperintensity volume, neuronal loss measured with Neurite Orientation Dispersion and Density Imaging, AD-specific atrophy patterns measured with the Spatial Patterns of Abnormality for Recognition of Early Alzheimer's Disease index, amyloid burden using amyloid positron emission tomography (PET), progression to mild cognitive impairment and baseline and longitudinal cognitive measures (trail making task, mini mental state examination, digit symbol substitution task).

FINDINGS

Individuals with advanced structural and functional brain-ages had more features indicative of neurodegeneration and they had poor cognition. Individuals with a resilient brain-age in both modalities had a genetic variant that has been shown to be associated with age of onset of AD. Mixed brain-age was associated with selective cognitive deficits.

INTERPRETATION

The advanced group displayed evidence of increased atrophy across all neuroimaging features that was not found in either of the mixed groups. This is in line with biomarkers of preclinical AD and cerebrovascular disease. These findings suggest that the variation in structural and functional brain ageing across individuals reflects the degree of underlying neuropathological processes and may indicate the propensity to develop dementia in later life.

FUNDING

The National Institute on Aging, the National Institutes of Health, the Swiss National Science Foundation, the Kaiser Foundation Research Institute and the National Heart, Lung, and Blood Institute.

ComBat models for harmonization of resting-state EEG features in multisite studies

OBJECTIVE

Pooling multisite resting-state electroencephalography (rsEEG) datasets may introduce bias due to batch effects (i.e., cross-site differences in the rsEEG related to scanner/sample characteristics). The Combining Batches (ComBat) models, introduced for microarray expression and adapted for neuroimaging, can control for batch effects while preserving the variability of biological covariates. We aim to evaluate four ComBat harmonization methods in a pooled sample from five independent rsEEG datasets of young and old adults.

METHODS

RsEEG signals (n = 374) were automatically preprocessed. Oscillatory and aperiodic rsEEG features were extracted in sensor space. Features were harmonized using neuroCombat (standard ComBat used in neuroimaging), neuroHarmonize (variant with nonlinear adjustment of covariates), OPNested-GMM (variant based on Gaussian Mixture Models to fit bimodal feature distributions), and HarmonizR (variant based on resampling to handle missing feature values). Relationships between rsEEG features and age were explored before and after harmonizing batch effects.

RESULTS

Batch effects were identified in rsEEG features. All ComBat methods reduced batch effects and features' dispersion; HarmonizR and OPNested-GMM ComBat achieved the greatest performance. Harmonized Beta power, individual Alpha peak frequency, Aperiodic exponent, and offset in posterior electrodes showed significant relations with age. All ComBat models maintained the direction of observed relationships while increasing the effect size.

CONCLUSIONS

ComBat models, particularly HarmonizeR and OPNested-GMM ComBat, effectively control for batch effects in rsEEG spectral features.

SIGNIFICANCE

This workflow can be used in multisite studies to harmonize batch effects in sensor-space rsEEG spectral features while preserving biological associations.

The status of MRI databases across the world focused on psychiatric and neurological disorders

Neuroimaging databases for neuro-psychiatric disorders enable researchers to implement data-driven research approaches by providing access to rich data that can be used to study disease, build and validate machine learning models, and even redefine disease spectra. The importance of sharing large, multi-center, multi-disorder databases has gradually been recognized in order to truly translate brain imaging knowledge into real-world clinical practice. Here, we review MRI databases that share data globally to serve multiple psychiatric or neurological disorders. We found 42 datasets consisting of 23,293 samples from patients with psychiatry and neurological disorders and healthy controls; 1245 samples from mood disorders (major depressive disorder and bipolar disorder), 2015 samples from developmental disorders (autism spectrum disorder, attention-deficit hyperactivity disorder), 675 samples from schizophrenia, 1194 samples from Parkinson's disease, 5865 samples from dementia (including Alzheimer's disease), We recognize that large, multi-center databases should include governance processes that allow data to be shared across national boundaries. Addressing technical and regulatory issues of existing databases can lead to better design and implementation and improve data access for the research community. The current trend toward the development of shareable MRI databases will contribute to a better understanding of the pathophysiology, diagnosis and assessment, and development of early interventions for neuropsychiatric disorders.

Modeling the impact of MRI acquisition bias on structural connectomes: Harmonizing structural connectomes

One way to increase the statistical power and generalizability of neuroimaging studies is to collect data at multiple sites or merge multiple cohorts. However, this usually comes with site-related biases due to the heterogeneity of scanners and acquisition parameters, negatively impacting sensitivity. Brain structural connectomes are not an exception: Being derived from T1-weighted and diffusion-weighted magnetic resonance images, structural connectivity is impacted by differences in imaging protocol. Beyond minimizing acquisition parameter differences, removing bias with postprocessing is essential. In this work we create, from the exhaustive Human Connectome Project Young Adult dataset, a resampled dataset of different b-values and spatial resolutions, modeling a cohort scanned across multiple sites. After demonstrating the statistical impact of acquisition parameters on connectivity, we propose a linear regression with explicit modeling of b-value and spatial resolution, and validate its performance on separate datasets. We show that b-value and spatial resolution affect connectivity in different ways and that acquisition bias can be reduced using a linear regression informed by the acquisition parameters while retaining interindividual differences and hence boosting fingerprinting performance. We also demonstrate the generative potential of our model, and its generalization capability in an independent dataset reflective of typical acquisition practices in clinical settings.

Generalizable Links Between Borderline Personality Traits and Functional Connectivity

BACKGROUND

Symptoms of borderline personality disorder (BPD) often manifest during adolescence, but the underlying relationship between these debilitating symptoms and the development of functional brain networks is not well understood. Here, we aimed to investigate how multivariate patterns of functional connectivity are associated with borderline personality traits in large samples of young adults and adolescents.

METHODS

We used functional magnetic resonance imaging data from young adults and adolescents from the HCP-YA (Human Connectome Project Young Adult) (n = 870, ages 22-37 years, 457 female) and the HCP-D (Human Connectome Project Development) (n = 223, ages 16-21 years, 121 female). A previously validated BPD proxy score was derived from the NEO Five-Factor Inventory. A ridge regression model with cross-validation and nested hyperparameter tuning was trained and tested in HCP-YA to predict BPD scores in unseen data from regional functional connectivity. The trained model was further tested on data from HCP-D without further tuning. Finally, we tested how the connectivity patterns associated with BPD aligned with age-related changes in connectivity.

RESULTS

Multivariate functional connectivity patterns significantly predicted out-of-sample BPD scores in unseen data in young adults (HCP-YA ppermuted = .001) and older adolescents (HCP-D ppermuted = .001). Regional predictive capacity was heterogeneous; the most predictive regions were found in functional systems relevant for emotion regulation and executive function, including the ventral attention network. Finally, regional functional connectivity patterns that predicted BPD scores aligned with those associated with development in youth.

CONCLUSIONS

Individual differences in functional connectivity in developmentally sensitive regions are associated with borderline personality traits.

Individualized Structural Perturbations on Normative Brain Connectome Restrict Deep Brain Stimulation Outcomes in Parkinson's Disease

BACKGROUND

Patients with Parkinson's disease (PD) respond to deep brain stimulation (DBS) variably. However, how brain substrates restrict DBS outcomes remains unclear.

OBJECTIVE

In this article, we aim to identify prognostic brain signatures for explaining the response variability.

METHODS

We retrospectively investigated a cohort of patients with PD (n = 141) between 2017 and 2022, and defined DBS outcomes as the improvement ratio of clinical motor scores. We used a deviation index to quantify individual perturbations on a reference structural covariance network acquired with preoperative T1-weighted magnetic resonance imaging. The neurobiological perturbations of patients were represented as z scored indices based on the chronological perturbations measured on a group of normal aging adults.

RESULTS

After applying stringent statistical tests (z > 2.5) and correcting for false discoveries (P < 0.01), we found that accelerated deviations mainly affected the prefrontal cortex, motor strip, limbic system, and cerebellum in PD. Particularly, a negative network within the accelerated deviations, expressed as "more preoperative deviations, less postoperative improvements," could predict DBS outcomes (mean absolute error = 0.09, R2 = 0.15). Moreover, a fusion of personal brain predictors and medical responses significantly improved traditional evaluations of DBS outcomes. Notably, the most important brain predictor, a pathway connecting the cognitive unit (prefrontal cortex) and motor control unit (cerebellum and motor strip), partially mediates DBS outcomes with the age at surgery.

CONCLUSIONS

Our findings suggest that individual structural perturbations on the cognitive motor control circuit are critical for modulating DBS outcomes. Interventions toward the circuit have the potential for additional clinical improvements. © 2024 The Author(s). Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

DeepComBat: A statistically motivated, hyperparameter-robust, deep learning approach to harmonization of neuroimaging data

Neuroimaging data acquired using multiple scanners or protocols are increasingly available. However, such data exhibit technical artifacts across batches which introduce confounding and decrease reproducibility. This is especially true when multi-batch data are analyzed using complex downstream models which are more likely to pick up on and implicitly incorporate batch-related information. Previously proposed image harmonization methods have sought to remove these batch effects; however, batch effects remain detectable in the data after applying these methods. We present DeepComBat, a deep learning harmonization method based on a conditional variational autoencoder and the ComBat method. DeepComBat combines the strengths of statistical and deep learning methods in order to account for the multivariate relationships between features while simultaneously relaxing strong assumptions made by previous deep learning harmonization methods. As a result, DeepComBat can perform multivariate harmonization while preserving data structure and avoiding the introduction of synthetic artifacts. We apply this method to cortical thickness measurements from a cognitive-aging cohort and show DeepComBat qualitatively and quantitatively outperforms existing methods in removing batch effects while preserving biological heterogeneity. Additionally, DeepComBat provides a new perspective for statistically motivated deep learning harmonization methods.

Brain-age prediction: Systematic evaluation of site effects, and sample age range and size

Structural neuroimaging data have been used to compute an estimate of the biological age of the brain (brain-age) which has been associated with other biologically and behaviorally meaningful measures of brain development and aging. The ongoing research interest in brain-age has highlighted the need for robust and publicly available brain-age models pre-trained on data from large samples of healthy individuals. To address this need we have previously released a developmental brain-age model. Here we expand this work to develop, empirically validate, and disseminate a pre-trained brain-age model to cover most of the human lifespan. To achieve this, we selected the best-performing model after systematically examining the impact of seven site harmonization strategies, age range, and sample size on brain-age prediction in a discovery sample of brain morphometric measures from 35,683 healthy individuals (age range: 5-90 years; 53.59% female). The pre-trained models were tested for cross-dataset generalizability in an independent sample comprising 2101 healthy individuals (age range: 8-80 years; 55.35% female) and for longitudinal consistency in a further sample comprising 377 healthy individuals (age range: 9-25 years; 49.87% female). This empirical examination yielded the following findings: (1) the accuracy of age prediction from morphometry data was higher when no site harmonization was applied; (2) dividing the discovery sample into two age-bins (5-40 and 40-90 years) provided a better balance between model accuracy and explained age variance than other alternatives; (3) model accuracy for brain-age prediction plateaued at a sample size exceeding 1600 participants. These findings have been incorporated into CentileBrain (https://centilebrain.org/#/brainAGE2), an open-science, web-based platform for individualized neuroimaging metrics.

Multivariate testing and effect size measures for batch effect evaluation in radiomic features

While precision medicine applications of radiomics analysis are promising, differences in image acquisition can cause "batch effects" that reduce reproducibility and affect downstream predictive analyses. Harmonization methods such as ComBat have been developed to correct these effects, but evaluation methods for quantifying batch effects are inconsistent. In this study, we propose the use of the multivariate statistical test PERMANOVA and the Robust Effect Size Index (RESI) to better quantify and characterize batch effects in radiomics data. We evaluate these methods in both simulated and real radiomics features extracted from full-field digital mammography (FFDM) data. PERMANOVA demonstrated higher power than standard univariate statistical testing, and RESI was able to interpretably quantify the effect size of site at extremely large sample sizes. These methods show promise as more powerful and interpretable methods for the detection and quantification of batch effects in radiomics studies.

SAN: Mitigating spatial covariance heterogeneity in cortical thickness data collected from multiple scanners or sites

In neuroimaging studies, combining data collected from multiple study sites or scanners is becoming common to increase the reproducibility of scientific discoveries. At the same time, unwanted variations arise by using different scanners (inter-scanner biases), which need to be corrected before downstream analyses to facilitate replicable research and prevent spurious findings. While statistical harmonization methods such as ComBat have become popular in mitigating inter-scanner biases in neuroimaging, recent methodological advances have shown that harmonizing heterogeneous covariances results in higher data quality. In vertex-level cortical thickness data, heterogeneity in spatial autocorrelation is a critical factor that affects covariance heterogeneity. Our work proposes a new statistical harmonization method called spatial autocorrelation normalization (SAN) that preserves homogeneous covariance vertex-level cortical thickness data across different scanners. We use an explicit Gaussian process to characterize scanner-invariant and scanner-specific variations to reconstruct spatially homogeneous data across scanners. SAN is computationally feasible, and it easily allows the integration of existing harmonization methods. We demonstrate the utility of the proposed method using cortical thickness data from the Social Processes Initiative in the Neurobiology of the Schizophrenia(s) (SPINS) study. SAN is publicly available as an R package.

Microstructural Mapping of Neural Pathways in Alzheimer's Disease using Macrostructure-Informed Normative Tractometry

Introduction

Diffusion MRI is sensitive to the microstructural properties of brain tissues, and shows great promise in detecting the effects of degenerative diseases. However, many approaches analyze single measures averaged over regions of interest, without considering the underlying fiber geometry.

Methods

Here, we propose a novel Macrostructure-Informed Normative Tractometry (MINT) framework, to investigate how white matter microstructure and macrostructure are jointly altered in mild cognitive impairment (MCI) and dementia. We compare MINT-derived metrics with univariate metrics from diffusion tensor imaging (DTI), to examine how fiber geometry may impact interpretation of microstructure.

Results

In two multi-site cohorts from North America and India, we find consistent patterns of microstructural and macrostructural anomalies implicated in MCI and dementia; we also rank diffusion metrics' sensitivity to dementia.

Discussion

We show that MINT, by jointly modeling tract shape and microstructure, has potential to disentangle and better interpret the effects of degenerative disease on the brain's neural pathways.

Functional connectivity development along the sensorimotor-association axis enhances the cortical hierarchy

Human cortical maturation has been posited to be organized along the sensorimotor-association axis, a hierarchical axis of brain organization that spans from unimodal sensorimotor cortices to transmodal association cortices. Here, we investigate the hypothesis that the development of functional connectivity during childhood through adolescence conforms to the cortical hierarchy defined by the sensorimotor-association axis. We tested this pre-registered hypothesis in four large-scale, independent datasets (total n = 3355; ages 5-23 years): the Philadelphia Neurodevelopmental Cohort (n = 1207), Nathan Kline Institute-Rockland Sample (n = 397), Human Connectome Project: Development (n = 625), and Healthy Brain Network (n = 1126). Across datasets, the development of functional connectivity systematically varied along the sensorimotor-association axis. Connectivity in sensorimotor regions increased, whereas connectivity in association cortices declined, refining and reinforcing the cortical hierarchy. These consistent and generalizable results establish that the sensorimotor-association axis of cortical organization encodes the dominant pattern of functional connectivity development.

Identifying Biases in a Multicenter MRI Database for Parkinson's Disease Classification: Is the Disease Classifier a Secret Site Classifier?

Sharing multicenter imaging datasets can be advantageous to increase data diversity and size but may lead to spurious correlations between site-related biological and non-biological image features and target labels, which machine learning (ML) models may exploit as shortcuts. To date, studies analyzing how and if deep learning models may use such effects as a shortcut are scarce. Thus, the aim of this work was to investigate if site-related effects are encoded in the feature space of an established deep learning model designed for Parkinson's disease (PD) classification based on T1-weighted MRI datasets. Therefore, all layers of the PD classifier were frozen, except for the last layer of the network, which was replaced by a linear layer that was exclusively re-trained to predict three potential bias types (biological sex, scanner type, and originating site). Our findings based on a large database consisting of 1880 MRI scans collected across 41 centers show that the feature space of the established PD model (74% accuracy) can be used to classify sex (75% accuracy), scanner type (79% accuracy), and site location (71% accuracy) with high accuracies despite this information never being explicitly provided to the PD model during original training. Overall, the results of this study suggest that trained image-based classifiers may use unwanted shortcuts that are not meaningful for the actual clinical task at hand. This finding may explain why many image-based deep learning models do not perform well when applied to data from centers not contributing to the training set.

SAN: mitigating spatial covariance heterogeneity in cortical thickness data collected from multiple scanners or sites

In neuroimaging studies, combining data collected from multiple study sites or scanners is becoming common to increase the reproducibility of scientific discoveries. At the same time, unwanted variations arise by using different scanners (inter-scanner biases), which need to be corrected before downstream analyses to facilitate replicable research and prevent spurious findings. While statistical harmonization methods such as ComBat have become popular in mitigating inter-scanner biases in neuroimaging, recent methodological advances have shown that harmonizing heterogeneous covariances results in higher data quality. In vertex-level cortical thickness data, heterogeneity in spatial autocorrelation is a critical factor that affects covariance heterogeneity. Our work proposes a new statistical harmonization method called SAN (Spatial Autocorrelation Normalization) that preserves homogeneous covariance vertex-level cortical thickness data across different scanners. We use an explicit Gaussian process to characterize scanner-invariant and scanner-specific variations to reconstruct spatially homogeneous data across scanners. SAN is computationally feasible, and it easily allows the integration of existing harmonization methods. We demonstrate the utility of the proposed method using cortical thickness data from the Social Processes Initiative in the Neurobiology of the Schizophrenia(s) (SPINS) study. SAN is publicly available as an R package.

Associations Between Structural Covariance Network and Antipsychotic Treatment Response in Schizophrenia

BACKGROUND AND HYPOTHESIS

Schizophrenia is associated with widespread cortical thinning and abnormality in the structural covariance network, which may reflect connectome alterations due to treatment effect or disease progression. Notably, patients with treatment-resistant schizophrenia (TRS) have stronger and more widespread cortical thinning, but it remains unclear whether structural covariance is associated with treatment response in schizophrenia.

STUDY DESIGN

We organized a multicenter magnetic resonance imaging study to assess structural covariance in a large population of TRS and non-TRS, who had been resistant and responsive to non-clozapine antipsychotics, respectively. Whole-brain structural covariance for cortical thickness was assessed in 102 patients with TRS, 77 patients with non-TRS, and 79 healthy controls (HC). Network-based statistics were used to examine the difference in structural covariance networks among the 3 groups. Moreover, the relationship between altered individual differentiated structural covariance and clinico-demographics was also explored.

STUDY RESULTS

Patients with non-TRS exhibited greater structural covariance compared with HC, mainly in the fronto-temporal and fronto-occipital regions, while there were no significant differences in structural covariance between TRS and non-TRS or HC. Higher individual differentiated structural covariance was associated with lower general scores of the Positive and Negative Syndrome Scale in the non-TRS group, but not in the TRS group.

CONCLUSIONS

These findings suggest that reconfiguration of brain networks via coordinated cortical thinning is related to treatment response in schizophrenia. Further longitudinal studies are warranted to confirm if greater structural covariance could serve as a marker for treatment response in this disease.

Disease Progression Patterns of Brain Morphology in Schizophrenia: More Progressed Stages in Treatment Resistance

BACKGROUND AND HYPOTHESIS

Given the heterogeneity and possible disease progression in schizophrenia, identifying the neurobiological subtypes and progression patterns in each patient may lead to novel biomarkers. Here, we adopted data-driven machine-learning techniques to identify the progression patterns of brain morphological changes in schizophrenia and investigate the association with treatment resistance.

STUDY DESIGN

In this cross-sectional multicenter study, we included 177 patients with schizophrenia, characterized by treatment response or resistance, with 3D T1-weighted magnetic resonance imaging. Cortical thickness and subcortical volumes calculated by FreeSurfer were converted into z scores using 73 healthy controls data. The Subtype and Stage Inference (SuStaIn) algorithm was used for unsupervised machine-learning analysis.

STUDY RESULTS

SuStaIn identified 3 different subtypes: (1) subcortical volume reduction (SC) type (73 patients), in which volume reduction of subcortical structures occurs first and moderate cortical thinning follows, (2) globus pallidus hypertrophy and cortical thinning (GP-CX) type (42 patients), in which globus pallidus hypertrophy initially occurs followed by progressive cortical thinning, and (3) cortical thinning (pure CX) type (39 patients), in which thinning of the insular and lateral temporal lobe cortices primarily happens. The remaining 23 patients were assigned to baseline stage of progression (no change). SuStaIn also found 84 stages of progression, and treatment-resistant schizophrenia showed significantly more progressed stages than treatment-responsive cases (P = .001). The GP-CX type presented earlier stages than the pure CX type (P = .009).

CONCLUSIONS

The brain morphological progressions in schizophrenia can be classified into 3 subtypes, and treatment resistance was associated with more progressed stages, which may suggest a novel biomarker.

Deep conditional generative model for longitudinal single-slice abdominal computed tomography harmonization

Purpose

Two-dimensional single-slice abdominal computed tomography (CT) provides a detailed tissue map with high resolution allowing quantitative characterization of relationships between health conditions and aging. However, longitudinal analysis of body composition changes using these scans is difficult due to positional variation between slices acquired in different years, which leads to different organs/tissues being captured.

Approach

To address this issue, we propose C-SliceGen, which takes an arbitrary axial slice in the abdominal region as a condition and generates a pre-defined vertebral level slice by estimating structural changes in the latent space.

Results

Our experiments on 2608 volumetric CT data from two in-house datasets and 50 subjects from the 2015 Multi-Atlas Abdomen Labeling Challenge Beyond the Cranial Vault (BTCV) dataset demonstrate that our model can generate high-quality images that are realistic and similar. We further evaluate our method's capability to harmonize longitudinal positional variation on 1033 subjects from the Baltimore longitudinal study of aging dataset, which contains longitudinal single abdominal slices, and confirmed that our method can harmonize the slice positional variance in terms of visceral fat area.

Conclusion

This approach provides a promising direction for mapping slices from different vertebral levels to a target slice and reducing positional variance for single-slice longitudinal analysis. The source code is available at: https://github.com/MASILab/C-SliceGen.

Lifespan reference curves for harmonizing multi-site regional brain white matter metrics from diffusion MRI

Age-related white matter (WM) microstructure maturation and decline occur throughout the human lifespan, complementing the process of gray matter development and degeneration. Here, we create normative lifespan reference curves for global and regional WM microstructure by harmonizing diffusion MRI (dMRI)-derived data from ten public datasets (N = 40,898 subjects; age: 3-95 years; 47.6% male). We tested three harmonization methods on regional diffusion tensor imaging (DTI) based fractional anisotropy (FA), a metric of WM microstructure, extracted using the ENIGMA-DTI pipeline. ComBat-GAM harmonization provided multi-study trajectories most consistent with known WM maturation peaks. Lifespan FA reference curves were validated with test-retest data and used to assess the effect of the ApoE4 risk factor for dementia in WM across the lifespan. We found significant associations between ApoE4 and FA in WM regions associated with neurodegenerative disease even in healthy individuals across the lifespan, with regional age-by-genotype interactions. Our lifespan reference curves and tools to harmonize new dMRI data to the curves are publicly available as eHarmonize (https://github.com/ahzhu/eharmonize).

Efficacy of MRI data harmonization in the age of machine learning: a multicenter study across 36 datasets

Pooling publicly-available MRI data from multiple sites allows to assemble extensive groups of subjects, increase statistical power, and promote data reuse with machine learning techniques. The harmonization of multicenter data is necessary to reduce the confounding effect associated with non-biological sources of variability in the data. However, when applied to the entire dataset before machine learning, the harmonization leads to data leakage, because information outside the training set may affect model building, and potentially falsely overestimate performance. We propose a 1) measurement of the efficacy of data harmonization; 2) harmonizer transformer, i.e., an implementation of the ComBat harmonization allowing its encapsulation among the preprocessing steps of a machine learning pipeline, avoiding data leakage by design. We tested these tools using brain T1-weighted MRI data from 1740 healthy subjects acquired at 36 sites. After harmonization, the site effect was removed or reduced, and we showed the data leakage effect in predicting individual age from MRI data, highlighting that introducing the harmonizer transformer into a machine learning pipeline allows for avoiding data leakage by design.

Neural signatures of emotion regulation

Emotional experience is central to a fulfilling life. Although exposure to negative experiences is inevitable, an individual's emotion regulation response may buffer against psychopathology. Identification of neural activation patterns associated with emotion regulation via an fMRI task is a promising and non-invasive means of furthering our understanding of the how the brain engages with negative experiences. Prior work has applied multivariate pattern analysis to identify signatures of response to negative emotion-inducing images; we adapt these techniques to establish novel neural signatures associated with conscious efforts to modulate emotional response. We model voxel-level activation via LASSO principal components regression and linear discriminant analysis to predict if a subject was engaged in emotion regulation and to identify brain regions which define this emotion regulation signature. We train our models using 82 participants and evaluate them on a holdout sample of 40 participants, demonstrating an accuracy up to 82.5% across three classes. Our results suggest that emotion regulation produces a unique signature that is differentiable from passive viewing of negative and neutral imagery.

Cardiorespiratory fitness is associated with cortical thickness of medial temporal brain areas associated with spatial cognition in young but not older adults

Cardiorespiratory fitness has a potent effect on neurocognitive health, especially regarding the hippocampal memory system. However, less is known about the impact of cardiorespiratory fitness on medial temporal lobe extrahippocampal neocortical regions. Specifically, it is unclear how cardiorespiratory fitness modulates these brain regions in young adulthood and if these regions are differentially related to cardiorespiratory fitness in young versus older adults. The primary goal of this study was to investigate if cardiorespiratory fitness predicted medial temporal lobe cortical thickness which, with the hippocampus, are critical for spatial learning and memory. Additionally, given the established role of these cortices in spatial navigation, we sought to determine if cardiorespiratory fitness and medial temporal lobe cortical thickness would predict greater subjective sense of direction in both young and older adults. Cross-sectional data from 56 young adults (20-35 years) and 44 older adults (55-85 years) were included. FreeSurfer 6.0 was used to automatically segment participants' 3T T1-weighted images. Using hierarchical multiple regression analyses, we confirmed significant associations between greater cardiorespiratory fitness and greater left entorhinal, left parahippocampal, and left perirhinal cortical thickness in young, but not older, adults. Left parahippocampal cortical thickness interacted with age group to differentially predict subjective sense of direction in young and older adults. Young adults displayed a positive, and older adults a negative, correlation between left parahippocampal cortical thickness and sense of direction. Our findings extend previous work on the association between cardiorespiratory fitness and hippocampal subfield structure in young adults to left medial temporal lobe neocortical regions.

Delineating a Pathway for the Discovery of Functional Connectome Biomarkers of Autism

The promise of individually tailored care for autism has driven efforts to establish biomarkers. This chapter appraises the state of precision-medicine research focused on biomarkers based on the functional brain connectome. This work is grounded on abundant evidence supporting the brain dysconnection model of autism and the advantages of resting-state functional MRI (R-fMRI) for studying the brain in vivo. After considering biomarker requirements of consistency and clinical relevance, we provide a scoping review of R-fMRI studies of individual prediction in autism. In the past 10 years, responding to the availability of open data through the Autism Brain Imaging Data Exchange, machine learning studies have surged. Nearly all have focused on diagnostic label classification. These efforts have shown that autism prediction is feasible using functional connectome markers, with accuracy reported well above chance. In parallel, emerging approaches more directly addressing autism heterogeneity are paving the way for much-needed biomarkers of longitudinal outcome and treatment response. We conclude with key challenges to be addressed by the next generation of studies.

Brain structural covariance network features are robust markers of early heavy alcohol use

BACKGROUND AND AIMS

Recently, we demonstrated that a distinct pattern of structural covariance networks (SCN) from magnetic resonance imaging (MRI)-derived measurements of brain cortical thickness characterized young adults with alcohol use disorder (AUD) and predicted current and future problematic drinking in adolescents relative to controls. Here, we establish the robustness and value of SCN for identifying heavy alcohol users in three additional independent studies.

DESIGN AND SETTING

Cross-sectional and longitudinal studies using data from the Pediatric Imaging, Neurocognition and Genetics (PING) study (n = 400, age range = 14-22 years), the National Consortium on Alcohol and Neurodevelopment in Adolescence (NCANDA) (n = 272, age range = 17-22 years) and the Human Connectome Project (HCP) (n = 375, age range = 22-37 years).

CASES

Cases were defined based on heavy alcohol use patterns or former alcohol use disorder (AUD) diagnoses: 50, 68 and 61 cases were identified. Controls had none or low alcohol use or absence of AUD: 350, 204 and 314 controls were selected.

MEASUREMENTS

Graph theory metrics of segregation and integration were used to summarize SCN.

FINDINGS

Mirroring our prior findings, and across the three data sets, cases had a lower clustering coefficient [area under the curve (AUC) = -0.029, P = 0.002], lower modularity (AUC = -0.14, P = 0.004), lower average shortest path length (AUC = -0.078, P = 0.017) and higher global efficiency (AUC = 0.007, P = 0.010). Local efficiency differences were marginal (AUC = -0.017, P = 0.052). That is, cases exhibited lower network segregation and higher integration, suggesting that adjacent nodes (i.e. brain regions) were less similar in thickness whereas spatially distant nodes were more similar.

CONCLUSION

Structural covariance network (SCN) differences in the brain appear to constitute an early marker of heavy alcohol use in three new data sets and, more generally, demonstrate the utility of SCN-derived metrics to detect brain-related psychopathology.

Source-based morphometry reveals structural brain pattern abnormalities in 22q11.2 deletion syndrome

22q11.2 deletion syndrome (22q11DS) is the most frequently occurring microdeletion in humans. It is associated with a significant impact on brain structure, including prominent reductions in gray matter volume (GMV), and neuropsychiatric manifestations, including cognitive impairment and psychosis. It is unclear whether GMV alterations in 22q11DS occur according to distinct structural patterns. Then, 783 participants (470 with 22q11DS: 51% females, mean age [SD] 18.2 [9.2]; and 313 typically developing [TD] controls: 46% females, mean age 18.0 [8.6]) from 13 datasets were included in the present study. We segmented structural T1-weighted brain MRI scans and extracted GMV images, which were then utilized in a novel source-based morphometry (SBM) pipeline (SS-Detect) to generate structural brain patterns (SBPs) that capture co-varying GMV. We investigated the impact of the 22q11.2 deletion, deletion size, intelligence quotient, and psychosis on the SBPs. Seventeen GMV-SBPs were derived, which provided spatial patterns of GMV covariance associated with a quantitative metric (i.e., loading score) for analysis. Patterns of topographically widespread differences in GMV covariance, including the cerebellum, discriminated individuals with 22q11DS from healthy controls. The spatial extents of the SBPs that revealed disparities between individuals with 22q11DS and controls were consistent with the findings of the univariate voxel-based morphometry analysis. Larger deletion size was associated with significantly lower GMV in frontal and occipital SBPs; however, history of psychosis did not show a strong relationship with these covariance patterns. 22q11DS is associated with distinct structural abnormalities captured by topographical GMV covariance patterns that include the cerebellum. Findings indicate that structural anomalies in 22q11DS manifest in a nonrandom manner and in distinct covarying anatomical patterns, rather than a diffuse global process. These SBP abnormalities converge with previously reported cortical surface area abnormalities, suggesting disturbances of early neurodevelopment as the most likely underlying mechanism.

Increasing hub disruption parallels dementia severity in autosomal dominant Alzheimer disease

Hub regions in the brain, recognized for their roles in ensuring efficient information transfer, are vulnerable to pathological alterations in neurodegenerative conditions, including Alzheimer Disease (AD). Given their essential role in neural communication, disruptions to these hubs have profound implications for overall brain network integrity and functionality. Hub disruption, or targeted impairment of functional connectivity at the hubs, is recognized in AD patients. Computational models paired with evidence from animal experiments hint at a mechanistic explanation, suggesting that these hubs may be preferentially targeted in neurodegeneration, due to their high neuronal activity levels-a phenomenon termed "activity-dependent degeneration". Yet, two critical issues were unresolved. First, past research hasn't definitively shown whether hub regions face a higher likelihood of impairment (targeted attack) compared to other regions or if impairment likelihood is uniformly distributed (random attack). Second, human studies offering support for activity-dependent explanations remain scarce. We applied a refined hub disruption index to determine the presence of targeted attacks in AD. Furthermore, we explored potential evidence for activity-dependent degeneration by evaluating if hub vulnerability is better explained by global connectivity or connectivity variations across functional systems, as well as comparing its timing relative to amyloid beta deposition in the brain. Our unique cohort of participants with autosomal dominant Alzheimer Disease (ADAD) allowed us to probe into the preclinical stages of AD to determine the hub disruption timeline in relation to expected symptom emergence. Our findings reveal a hub disruption pattern in ADAD aligned with targeted attacks, detectable even in pre-clinical stages. Notably, the disruption's severity amplified alongside symptomatic progression. Moreover, since excessive local neuronal activity has been shown to increase amyloid deposition and high connectivity regions show high level of neuronal activity, our observation that hub disruption was primarily tied to regional differences in global connectivity and sequentially followed changes observed in Aβ PET cortical markers is consistent with the activity-dependent degeneration model. Intriguingly, these disruptions were discernible 8 years before the expected age of symptom onset. Taken together, our findings not only align with the targeted attack on hubs model but also suggest that activity-dependent degeneration might be the cause of hub vulnerability. This deepened understanding could be instrumental in refining diagnostic techniques and developing targeted therapeutic strategies for AD in the future.

Generalizable links between symptoms of borderline personality disorder and functional connectivity

Background |

Symptoms of borderline personality disorder (BPD) often manifest in adolescence, yet the underlying relationship between these debilitating symptoms and the development of functional brain networks is not well understood. Here we aimed to investigate how multivariate patterns of functional connectivity are associated with symptoms of BPD in a large sample of young adults and adolescents.

Methods |

We used high-quality functional Magnetic Resonance Imaging (fMRI) data from young adults from the Human Connectome Project: Young Adults (HCP-YA; N = 870, ages 22-37 years, 457 female) and youth from the Human Connectome Project: Development (HCP-D; N = 223, age range 16-21 years, 121 female). A previously validated BPD proxy score was derived from the NEO Five Factor Inventory (NEO-FFI). A ridge regression model with 10-fold cross-validation and nested hyperparameter tuning was trained and tested in HCP-YA to predict BPD scores in unseen data from regional functional connectivity, while controlling for in-scanner motion, age, and sex. The trained model was further tested on data from HCP-D without further tuning. Finally, we tested how the connectivity patterns associated with BPD aligned with age-related changes in connectivity.

Results |

Multivariate functional connectivity patterns significantly predicted out-of-sample BPD proxy scores in unseen data in both young adults (HCP-YA; pperm = 0.001) and older adolescents (HCP-D; pperm = 0.001). Predictive capacity of regions was heterogeneous; the most predictive regions were found in functional systems relevant for emotion regulation and executive function, including the ventral attention network. Finally, regional functional connectivity patterns that predicted BPD proxy scores aligned with those associated with development in youth.

Conclusion |

Individual differences in functional connectivity in developmentally-sensitive regions are associated with the symptoms of BPD.

RELIEF: A structured multivariate approach for removal of latent inter-scanner effects

Combining data collected from multiple study sites is becoming common and is advantageous to researchers to increase the generalizability and replicability of scientific discoveries. However, at the same time, unwanted inter-scanner biases are commonly observed across neuroimaging data collected from multiple study sites or scanners, rendering difficulties in integrating such data to obtain reliable findings. While several methods for handling such unwanted variations have been proposed, most of them use univariate approaches that could be too simple to capture all sources of scanner-specific variations. To address these challenges, we propose a novel multivariate harmonization method called RELIEF (REmoval of Latent Inter-scanner Effects through Factorization) for estimating and removing both explicit and latent scanner effects. Our method is the first approach to introduce the simultaneous dimension reduction and factorization of interlinked matrices to a data harmonization context, which provides a new direction in methodological research for correcting inter-scanner biases. Analyzing diffusion tensor imaging (DTI) data from the Social Processes Initiative in Neurobiology of the Schizophrenia (SPINS) study and conducting extensive simulation studies, we show that RELIEF outperforms existing harmonization methods in mitigating inter-scanner biases and retaining biological associations of interest to increase statistical power. RELIEF is publicly available as an R package.

Personalized estimates of brain cortical structural variability in individuals with Autism spectrum disorder: the predictor of brain age and neurobiology relevance

BACKGROUND

Autism spectrum disorder (ASD) is a heritable condition related to brain development that affects a person's perception and socialization with others. Here, we examined variability in the brain morphology in ASD children and adolescent individuals at the level of brain cortical structural profiles and the level of each brain regional measure.

METHODS

We selected brain structural MRI data in 600 ASDs and 729 normal controls (NCs) from Autism Brain Imaging Data Exchange (ABIDE). The personalized estimate of similarity between gray matter volume (GMV) profiles of an individual to that of others in the same group was assessed by using the person-based similarity index (PBSI). Regional contributions to PBSI score were utilized for brain age gap estimation (BrainAGE) prediction model establishment, including support vector regression (SVR), relevance vector regression (RVR), and Gaussian process regression (GPR). The association between BrainAGE prediction in ASD and clinical performance was investigated. We further explored the related inter-regional profiles of gene expression from the Allen Human Brain Atlas with variability differences in the brain morphology between groups.

RESULTS

The PBSI score of GMV was negatively related to age regardless of the sample group, and the PBSI score was significantly lower in ASDs than in NCs. The regional contributions to the PBSI score of 126 brain regions in ASDs showed significant differences compared to NCs. RVR model achieved the best performance for predicting brain age. Higher inter-individual brain morphology variability was related to increased brain age, specific to communication symptoms. A total of 430 genes belonging to various pathways were identified as associated with brain cortical morphometric variation. The pathways, including short-term memory, regulation of system process, and regulation of nervous system process, were dominated mainly by gene sets for manno midbrain neurotypes.

LIMITATIONS

There is a sample mismatch between the gene expression data and brain imaging data from ABIDE. A larger sample size can contribute to the model training of BrainAGE and the validation of the results.

CONCLUSIONS

ASD has personalized heterogeneity brain morphology. The brain age gap estimation and transcription-neuroimaging associations derived from this trait are replenished in an additional direction to boost the understanding of the ASD brain.

Image harmonization: A review of statistical and deep learning methods for removing batch effects and evaluation metrics for effective harmonization

Magnetic resonance imaging and computed tomography from multiple batches (e.g. sites, scanners, datasets, etc.) are increasingly used alongside complex downstream analyses to obtain new insights into the human brain. However, significant confounding due to batch-related technical variation, called batch effects, is present in this data; direct application of downstream analyses to the data may lead to biased results. Image harmonization methods seek to remove these batch effects and enable increased generalizability and reproducibility of downstream results. In this review, we describe and categorize current approaches in statistical and deep learning harmonization methods. We also describe current evaluation metrics used to assess harmonization methods and provide a standardized framework to evaluate newly-proposed methods for effective harmonization and preservation of biological information. Finally, we provide recommendations to end-users to advocate for more effective use of current methods and to methodologists to direct future efforts and accelerate development of the field.

Functional connectivity signatures of major depressive disorder: machine learning analysis of two multicenter neuroimaging studies

The promise of machine learning has fueled the hope for developing diagnostic tools for psychiatry. Initial studies showed high accuracy for the identification of major depressive disorder (MDD) with resting-state connectivity, but progress has been hampered by the absence of large datasets. Here we used regular machine learning and advanced deep learning algorithms to differentiate patients with MDD from healthy controls and identify neurophysiological signatures of depression in two of the largest resting-state datasets for MDD. We obtained resting-state functional magnetic resonance imaging data from the REST-meta-MDD (N = 2338) and PsyMRI (N = 1039) consortia. Classification of functional connectivity matrices was done using support vector machines (SVM) and graph convolutional neural networks (GCN), and performance was evaluated using 5-fold cross-validation. Features were visualized using GCN-Explainer, an ablation study and univariate t-testing. The results showed a mean classification accuracy of 61% for MDD versus controls. Mean accuracy for classifying (non-)medicated subgroups was 62%. Sex classification accuracy was substantially better across datasets (73-81%). Visualization of the results showed that classifications were driven by stronger thalamic connections in both datasets, while nearly all other connections were weaker with small univariate effect sizes. These results suggest that whole brain resting-state connectivity is a reliable though poor biomarker for MDD, presumably due to disease heterogeneity as further supported by the higher accuracy for sex classification using the same methods. Deep learning revealed thalamic hyperconnectivity as a prominent neurophysiological signature of depression in both multicenter studies, which may guide the development of biomarkers in future studies.

Reconfiguration of brain network dynamics underlying spatial deficits in subjective cognitive decline

Brain dynamics and the associations with spatial navigation in individuals with subjective cognitive decline (SCD) remain unknown. In this study, a hidden Markov model (HMM) was inferred from resting-state functional magnetic resonance imaging data in a cohort of 80 SCD and 77 normal control (NC) participants. By HMM, 12 states with distinct brain activity were identified. The SCD group showed increased fractional occupancy in the states with less activated ventral default mode, posterior salience, and visuospatial networks, while decreased fractional occupancy in the state with general network activation. The SCD group also showed decreased probabilities of transition into and out of the state with general network activation, suggesting an inability to dynamically upregulate and downregulate brain network activity. Significant correlations between brain dynamics and spatial navigation were observed. The combined features of spatial navigation and brain dynamics showed an area under the curve of 0.854 in distinguishing between SCD and NC. The findings may provide exploratory evidence of the reconfiguration of brain network dynamics underlying spatial deficits in SCD.

Harmonizing Healthy Cohorts to Support Multicenter Studies on Migraine Classification using Brain MRI Data

Multicenter and multi-scanner imaging studies might be needed to provide sample sizes large enough for developing accurate predictive models. However, multicenter studies, which likely include confounding factors due to subtle differences in research participant characteristics, MRI scanners, and imaging acquisition protocols, might not yield generalizable machine learning models, that is, models developed using one dataset may not be applicable to a different dataset. The generalizability of classification models is key for multi-scanner and multicenter studies, and for providing reproducible results. This study developed a data harmonization strategy to identify healthy controls with similar (homogenous) characteristics from multicenter studies to validate the generalization of machine-learning techniques for classifying individual migraine patients and healthy controls using brain MRI data. The Maximum Mean Discrepancy (MMD) was used to compare the two datasets represented in Geodesic Flow Kernel (GFK) space, capturing the data variabilities for identifying a "healthy core". A set of homogeneous healthy controls can assist in overcoming some of the unwanted heterogeneity and allow for the development of classification models that have high accuracy when applied to new datasets. Extensive experimental results show the utilization of a healthy core. One dataset consists of 120 individuals (66 with migraine and 54 healthy controls) and another dataset consists of 76 (34 with migraine and 42 healthy controls) individuals. A homogeneous dataset derived from a cohort of healthy controls improves the performance of classification models by about 25% accuracy improvements for both episodic and chronic migraineurs.

DeepComBat: A Statistically Motivated, Hyperparameter-Robust, Deep Learning Approach to Harmonization of Neuroimaging Data

Neuroimaging data from multiple batches (i.e. acquisition sites, scanner manufacturer, datasets, etc.) are increasingly necessary to gain new insights into the human brain. However, multi-batch data, as well as extracted radiomic features, exhibit pronounced technical artifacts across batches. These batch effects introduce confounding into the data and can obscure biological effects of interest, decreasing the generalizability and reproducibility of findings. This is especially true when multi-batch data is used alongside complex downstream analysis models, such as machine learning methods. Image harmonization methods seeking to remove these batch effects are important for mitigating these issues; however, significant multivariate batch effects remain in the data following harmonization by current state-of-the-art statistical and deep learning methods. We present DeepCombat, a deep learning harmonization method based on a conditional variational autoencoder architecture and the ComBat harmonization model. DeepCombat learns and removes subject-level batch effects by accounting for the multivariate relationships between features. Additionally, DeepComBat relaxes a number of strong assumptions commonly made by previous deep learning harmonization methods and is empirically robust across a wide range of hyperparameter choices. We apply this method to neuroimaging data from a large cognitive-aging cohort and find that DeepCombat outperforms existing methods, as assessed by a battery of machine learning methods, in removing scanner effects from cortical thickness measurements while preserving biological heterogeneity. Additionally, DeepComBat provides a new perspective for statistically-motivated deep learning harmonization methods.

Comprehensive evaluation of harmonization on functional brain imaging for multisite data-fusion

To embrace big-data neuroimaging, harmonizing the site effect in resting-state functional magnetic resonance imaging (R-fMRI) data fusion is a fundamental challenge. A comprehensive evaluation of potentially effective harmonization strategies, particularly with specifically collected data, has been scarce, especially for R-fMRI metrics. Here, we comprehensively assess harmonization strategies from multiple perspectives, including tests on residual site effect, individual identification, test-retest reliability, and replicability of group-level statistical results, on widely used R-fMRI metrics across various datasets, including data obtained from participants with repetitive measures at different scanners. For individual identifiability (i.e., whether the same subject could be identified across R-fMRI data scanned across different sites), we found that, while most methods decreased site effects, the Subsampling Maximum-mean-distance based distribution shift correction Algorithm (SMA) and parametric unadjusted CovBat outperformed linear regression models, linear mixed models, ComBat series and invariant conditional variational auto-encoder in clustering accuracy. Test-retest reliability was better for SMA and parametric adjusted CovBat than unadjusted ComBat series and parametric unadjusted CovBat in the number of overlapped voxels. At the same time, SMA was superior to the latter in replicability in terms of the Dice coefficient and the scale of brain areas showing sex differences reproducibly observed across datasets. Furthermore, SMA better detected reproducible sex differences of ALFF under the site-sex confounded situation. Moreover, we designed experiments to identify the best target site features to optimize SMA identifiability, test-retest reliability, and stability. We noted both sample size and distribution of the target site matter and introduced a heuristic formula for selecting the target site. In addition to providing practical guidelines, this work can inform continuing improvements and innovations in harmonizing methodologies for big R-fMRI data.