Toward a unified framework for interpreting machine-learning models in neuroimaging

Iowa Brain-Behavior Modeling Toolkit: An Open-Source MATLAB Tool for Inferential and Predictive Modeling of Imaging-Behavior and Lesion-Deficit Relationships

The traditional analytical framework taken by neuroimaging studies in general, and lesion-behavior studies in particular, has been inferential in nature and has focused on identifying and interpreting statistically significant effects within the sample under study. While this framework is well-suited for hypothesis testing approaches, achieving the modern goal of precision medicine requires a different framework that is predictive in nature and that focuses on maximizing the predictive power of models and evaluating their ability to generalize beyond the data that were used to train them. However, few tools exist to support the development and evaluation of predictive models in the context of neuroimaging or lesion-behavior research, creating an obstacle to the widespread adoption of predictive modeling approaches in the field. Further, existing tools for lesion-behavior analysis are often unable to accommodate categorical outcome variables and often impose restrictions on the predictor data. Researchers therefore often must use different software packages and analytical approaches depending on whether they are addressing a classification vs. regression problem and on whether their predictor data correspond to binary lesion images, continuous lesion-network images, connectivity matrices, or other data modalities. To address these limitations, we have developed a MATLAB software toolkit that supports both inferential and predictive modeling frameworks, accommodates both classification and regression problems, and does not impose restrictions on the modality of the predictor data. The toolkit features both a graphical user interface and scripting interface, includes implementations of multiple mass-univariate, multivariate, and machine learning models, features built-in and customizable routines for hyper-parameter optimization, cross-validation, model stacking, and significance testing, and automatically generates text-based descriptions of key methodological details and modeling results to improve reproducibility and minimize errors in the reporting of methods and results. Here, we provide an overview and discussion of the toolkit features and demonstrate its functionality by applying it to the question of how expressive and receptive language impairments relate to lesion location, structural disconnection, and functional network disruption in a large sample of patients with left hemispheric brain lesions. We find that impairments in expressive vs. receptive language are most strongly associated with left lateral prefrontal and left posterior temporal/parietal damage, respectively. We also find that impairments in expressive vs. receptive language are associated with partially overlapping patterns of fronto-temporal structural disconnection, and that the associated functional networks are also similar. Importantly, we find that lesion location and lesion-derived network measures are highly predictive of both types of impairment, with predictions from models trained on these measures explaining ~30-40% of the variance on average when applied to data from patients not used to train the models. We have made the toolkit publicly available, and we have included a comprehensive set of tutorial notebooks to support new users in applying the toolkit in their studies.

Brain-phenotype predictions of language and executive function can survive across diverse real-world data: Dataset shifts in developmental populations

Predictive modeling potentially increases the reproducibility and generalizability of neuroimaging brain-phenotype associations. Yet, the evaluation of a model in another dataset is underutilized. Among studies that undertake external validation, there is a notable lack of attention to generalization across dataset-specific idiosyncrasies (i.e., dataset shifts). Research settings, by design, remove the between-site variations that real-world and, eventually, clinical applications demand. Here, we rigorously test the ability of a range of predictive models to generalize across three diverse, unharmonized developmental samples: the Philadelphia Neurodevelopmental Cohort (n=1291), the Healthy Brain Network (n=1110), and the Human Connectome Project in Development (n=428). These datasets have high inter-dataset heterogeneity, encompassing substantial variations in age distribution, sex, racial and ethnic minority representation, recruitment geography, clinical symptom burdens, fMRI tasks, sequences, and behavioral measures. Through advanced methodological approaches, we demonstrate that reproducible and generalizable brain-behavior associations can be realized across diverse dataset features. Results indicate the potential of functional connectome-based predictive models to be robust despite substantial inter-dataset variability. Notably, for the HCPD and HBN datasets, the best predictions were not from training and testing in the same dataset (i.e., cross-validation) but across datasets. This result suggests that training on diverse data may improve prediction in specific cases. Overall, this work provides a critical foundation for future work evaluating the generalizability of brain-phenotype associations in real-world scenarios and clinical settings.

Advances and challenges in neuroimaging-based pain biomarkers

Identifying neural biomarkers of pain has long been a central theme in pain neuroscience. Here, we review the state-of-the-art candidates for neural biomarkers of acute and chronic pain. We classify these potential neural biomarkers into five categories based on the nature of their target variables, including neural biomarkers of (1) within-individual perception, (2) between-individual sensitivity, and (3) discriminability for acute pain, as well as (4) assessment and (5) prospective neural biomarkers for chronic pain. For each category, we provide a synthesized review of candidate biomarkers developed using neuroimaging techniques including functional magnetic resonance imaging (fMRI), structural magnetic resonance imaging (sMRI), and electroencephalography (EEG). We also discuss the conceptual and practical challenges in developing neural biomarkers of pain. Addressing these challenges, optimal biomarkers of pain can be developed to deepen our understanding of how the brain represents pain and ultimately help alleviate patients' suffering and improve their well-being.

The Neurocomputational Mechanism Underlying Decision-Making on Unfairness to Self and Others

Fairness is a fundamental value in human societies, with individuals concerned about unfairness both to themselves and to others. Nevertheless, an enduring debate focuses on whether self-unfairness and other-unfairness elicit shared or distinct neuropsychological processes. To address this, we combined a three-person ultimatum game with computational modeling and advanced neuroimaging analysis techniques to unravel the behavioral, cognitive, and neural patterns underlying unfairness to self and others. Our behavioral and computational results reveal a heightened concern among participants for self-unfairness over other-unfairness. Moreover, self-unfairness consistently activates brain regions such as the anterior insula, dorsal anterior cingulate cortex, and dorsolateral prefrontal cortex, spanning various spatial scales that encompass univariate activation, local multivariate patterns, and whole-brain multivariate patterns. These regions are well-established in their association with emotional and cognitive processes relevant to fairness-based decision-making. Conversely, other-unfairness primarily engages the middle occipital gyrus. Collectively, our findings robustly support distinct neurocomputational signatures between self-unfairness and other-unfairness.

Mental state decoders: game-changers or wishful thinking?

Decoding mental and perceptual states using fMRI has become increasingly popular over the past two decades, with numerous highly-cited studies published in high-profile journals. Nevertheless, what have we learned from these decoders? In this opinion, we argue that fMRI-based decoders are not neurophysiologically informative and are not, and likely cannot be, applicable to real-world decision-making. The former point stems from the fact that decoding models cannot disentangle neural mechanisms from their epiphenomena. The latter point stems from both logical and ethical constraints. Constructing decoders requires precious time and resources that should instead be directed toward scientific endeavors more likely to yield meaningful scientific progress.

Machine learning identifies different related factors associated with depression and suicidal ideation in Chinese children and adolescents

BACKGROUND

Depression and suicidal ideation often co-occur in children and adolescents, yet they possess distinct characteristics. This study sought to identify the different related factors associated with depression and suicidal ideation.

METHODS

A nationwide cross-sectional survey collected data from Chinese children and adolescents aged 8 to 18 (N = 160,962; 48.91 % girls). The survey included inquiries about demographics, depression, suicidal ideation, anxiety, perceived stress, academic burnout, internet addiction, non-suicidal self-injury, bullying, and being bullied. Fifteen machine learning algorithms were conducted to identify the different related factors associated with depression and suicidal ideation. Additionally, we conducted external validation on an independent sample of 1,812,889 children and adolescents.

RESULTS

Our findings revealed seven related factors linked to depression and six associated with suicidal ideation, with average accuracy rates of 86.86 % and 85.82 %, respectively. For depression, the most influential factors were anxiety, perceived stress, academic burnout, internet addiction, non-suicidal self-injury, experience of bullying, and age. Similarly, anxiety, non-suicidal self-injury, perceived stress, internet addiction, academic burnout, and age emerged as paramount factors for suicidal ideation. Moreover, these related factors showed notable variations in their predictive capacities for depression and suicidal ideation across different subgroups.

CONCLUSION

Anxiety emerged as the predominant shared factor for both depression and suicidal ideation, whereas the other related factors displayed distinct predictive patterns for each condition. These findings highlight the critical need for tailored strategies from public mental health service providers and policymakers to address the pressing concerns of depression and suicidal ideation.

Houston, We Have AI Problem! Quality Issues with Neuroimaging-Based Artificial Intelligence in Parkinson's Disease: A Systematic Review

In recent years, many neuroimaging studies have applied artificial intelligence (AI) to facilitate existing challenges in Parkinson's disease (PD) diagnosis, prognosis, and intervention. The aim of this systematic review was to provide an overview of neuroimaging-based AI studies and to assess their methodological quality. A PubMed search yielded 810 studies, of which 244 that investigated the utility of neuroimaging-based AI for PD diagnosis, prognosis, or intervention were included. We systematically categorized studies by outcomes and rated them with respect to five minimal quality criteria (MQC) pertaining to data splitting, data leakage, model complexity, performance reporting, and indication of biological plausibility. We found that the majority of studies aimed to distinguish PD patients from healthy controls (54%) or atypical parkinsonian syndromes (25%), whereas prognostic or interventional studies were sparse. Only 20% of evaluated studies passed all five MQC, with data leakage, non-minimal model complexity, and reporting of biological plausibility as the primary factors for quality loss. Data leakage was associated with a significant inflation of accuracies. Very few studies employed external test sets (8%), where accuracy was significantly lower, and 19% of studies did not account for data imbalance. Adherence to MQC was low across all observed years and journal impact factors. This review outlines that AI has been applied to a wide variety of research questions pertaining to PD; however, the number of studies failing to pass the MQC is alarming. Therefore, we provide recommendations to enhance the interpretability, generalizability, and clinical utility of future AI applications using neuroimaging in PD. © 2024 The Author(s). Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

Are deep learning classification results obtained on CT scans fair and interpretable?

Following the great success of various deep learning methods in image and object classification, the biomedical image processing society is also overwhelmed with their applications to various automatic diagnosis cases. Unfortunately, most of the deep learning-based classification attempts in the literature solely focus on the aim of extreme accuracy scores, without considering interpretability, or patient-wise separation of training and test data. For example, most lung nodule classification papers using deep learning randomly shuffle data and split it into training, validation, and test sets, causing certain images from the Computed Tomography (CT) scan of a person to be in the training set, while other images of the same person to be in the validation or testing image sets. This can result in reporting misleading accuracy rates and the learning of irrelevant features, ultimately reducing the real-life usability of these models. When the deep neural networks trained on the traditional, unfair data shuffling method are challenged with new patient images, it is observed that the trained models perform poorly. In contrast, deep neural networks trained with strict patient-level separation maintain their accuracy rates even when new patient images are tested. Heat map visualizations of the activations of the deep neural networks trained with strict patient-level separation indicate a higher degree of focus on the relevant nodules. We argue that the research question posed in the title has a positive answer only if the deep neural networks are trained with images of patients that are strictly isolated from the validation and testing patient sets.

Progress and trends in neurological disorders research based on deep learning

In recent years, deep learning (DL) has emerged as a powerful tool in clinical imaging, offering unprecedented opportunities for the diagnosis and treatment of neurological disorders (NDs). This comprehensive review explores the multifaceted role of DL techniques in leveraging vast datasets to advance our understanding of NDs and improve clinical outcomes. Beginning with a systematic literature review, we delve into the utilization of DL, particularly focusing on multimodal neuroimaging data analysis-a domain that has witnessed rapid progress and garnered significant scientific interest. Our study categorizes and critically analyses numerous DL models, including Convolutional Neural Networks (CNNs), LSTM-CNN, GAN, and VGG, to understand their performance across different types of Neurology Diseases. Through particular analysis, we identify key benchmarks and datasets utilized in training and testing DL models, shedding light on the challenges and opportunities in clinical neuroimaging research. Moreover, we discuss the effectiveness of DL in real-world clinical scenarios, emphasizing its potential to revolutionize ND diagnosis and therapy. By synthesizing existing literature and describing future directions, this review not only provides insights into the current state of DL applications in ND analysis but also covers the way for the development of more efficient and accessible DL techniques. Finally, our findings underscore the transformative impact of DL in reshaping the landscape of clinical neuroimaging, offering hope for enhanced patient care and groundbreaking discoveries in the field of neurology. This review paper is beneficial for neuropathologists and new researchers in this field.

Seeing more than the Tip of the Iceberg: Approaches to Subthreshold Effects in Functional Magnetic Resonance Imaging of the Brain

Many functional magnetic resonance imaging (fMRI) studies and presurgical mapping applications rely on mass-univariate inference with subsequent multiple comparison correction. Statistical results are frequently visualized as thresholded statistical maps. This approach has inherent limitations including the risk of drawing overly-selective conclusions based only on selective results passing such thresholds. This article gives an overview of both established and newly emerging scientific approaches to supplement such conventional analyses by incorporating information about subthreshold effects with the aim to improve interpretation of findings or leverage a wider array of information. Topics covered include neuroimaging data visualization, p-value histogram analysis and the related Higher Criticism approach for detecting rare and weak effects. Further examples from multivariate analyses and dedicated Bayesian approaches are provided.

Identifying gray matter alterations in Cushing's disease using machine learning: An interpretable approach

BACKGROUND

Cushing's Disease (CD) is a rare clinical syndrome characterized by excessive secretion of adrenocorticotrophic hormone, leading to significant functional and structural brain alterations as observed in Magnetic Resonance Imaging (MRI). While traditional statistical analysis has been widely employed to investigate these MRI changes in CD, it has lacked the ability to predict individual-level outcomes.

PURPOSE

To address this problem, this paper has proposed an interpretable machine learning (ML) framework, including model-level assessment, feature-level assessment, and biology-level assessment to ensure a comprehensive analysis based on structural MRI of CD.

METHODS

The ML framework has effectively identified the changes in brain regions in the stage of model-level assessment, verified the effectiveness of these altered brain regions to predict CD from normal controls in the stage of feature-level assessment, and carried out a correlation analysis between altered brain regions and clinical symptoms in the stage of biology-level assessment.

RESULTS

The experimental results of this study have demonstrated that the Insula, Fusiform gyrus, Superior frontal gyrus, Precuneus, and the opercular portion of the Inferior frontal gyrus of CD showed significant alterations in brain regions. Furthermore, our study has revealed significant correlations between clinical symptoms and the frontotemporal lobes, insulin, and olfactory cortex, which also have been confirmed by previous studies.

CONCLUSIONS

The ML framework proposed in this study exhibits exceptional potential in uncovering the intricate pathophysiological mechanisms underlying CD, with potential applicability in diagnosing other diseases.

A neurofunctional signature of subjective disgust generalizes to oral distaste and socio-moral contexts

While disgust originates in the hard-wired mammalian distaste response, the conscious experience of disgust in humans strongly depends on subjective appraisal and may even extend to socio-moral contexts. Here, in a series of studies, we combined functional magnetic resonance imaging with machine-learning-based predictive modelling to establish a comprehensive neurobiological model of subjective disgust. The developed neurofunctional signature accurately predicted momentary self-reported subjective disgust across discovery (n = 78) and pre-registered validation (n = 30) cohorts and generalized across core disgust (n = 34 and n = 26), gustatory distaste (n = 30) and socio-moral (unfair offers; n = 43) contexts. Disgust experience was encoded in distributed cortical and subcortical systems, and exhibited distinct and shared neural representations with subjective fear or negative affect in interoceptive-emotional awareness and conscious appraisal systems, while the signatures most accurately predicted the respective target experience. We provide an accurate functional magnetic resonance imaging signature for disgust with a high potential to resolve ongoing evolutionary debates.

Intracranial EEG signals disentangle multi-areal neural dynamics of vicarious pain perception

Empathy enables understanding and sharing of others' feelings. Human neuroimaging studies have identified critical brain regions supporting empathy for pain, including the anterior insula (AI), anterior cingulate (ACC), amygdala, and inferior frontal gyrus (IFG). However, to date, the precise spatio-temporal profiles of empathic neural responses and inter-regional communications remain elusive. Here, using intracranial electroencephalography, we investigated electrophysiological signatures of vicarious pain perception. Others' pain perception induced early increases in high-gamma activity in IFG, beta power increases in ACC, but decreased beta power in AI and amygdala. Vicarious pain perception also altered the beta-band-coordinated coupling between ACC, AI, and amygdala, as well as increased modulation of IFG high-gamma amplitudes by beta phases of amygdala/AI/ACC. We identified a necessary combination of neural features for decoding vicarious pain perception. These spatio-temporally specific regional activities and inter-regional interactions within the empathy network suggest a neurodynamic model of human pain empathy.

Stable multivariate lesion symptom mapping

Multivariate lesion-symptom mapping (MLSM) considers lesion information across the entire brain to predict impairments. The strength of this approach is also its weakness-considering many brain features together synergistically can uncover complex brain-behavior relationships but exposes a high-dimensional feature space that a model is expected to learn. Successfully distinguishing between features in this landscape can be difficult for models, particularly in the presence of irrelevant or redundant features. Here, we propose stable multivariate lesion-symptom mapping (sMLSM), which integrates the identification of reliable features with stability selection into conventional MLSM and describe our open-source MATLAB implementation. Usage is showcased with our publicly available dataset of chronic stroke survivors (N=167) and further validated in our independent public acute stroke dataset (N = 1106). We demonstrate that sMLSM eliminates inconsistent features highlighted by MLSM, reduces variation in feature weights, enables the model to learn more complex patterns of brain damage, and improves model accuracy for predicting aphasia severity in a way that tends to be robust regarding the choice of parameters for identifying reliable features. Critically, sMLSM more consistently outperforms predictions based on lesion size alone. This advantage is evident starting at modest sample sizes (N>75). Spatial distribution of feature importance is different in sMLSM, which highlights the features identified by univariate lesion symptom mapping while also implicating select regions emphasized by MLSM. Beyond improved prediction accuracy, sMLSM can offer deeper insight into reliable biomarkers of impairment, informing our understanding of neurobiology.

TractGeoNet: A geometric deep learning framework for pointwise analysis of tract microstructure to predict language assessment performance

We propose a geometric deep-learning-based framework, TractGeoNet, for performing regression using diffusion magnetic resonance imaging (dMRI) tractography and associated pointwise tissue microstructure measurements. By employing a point cloud representation, TractGeoNet can directly utilize tissue microstructure and positional information from all points within a fiber tract without the need to average or bin data along the streamline as traditionally required by dMRI tractometry methods. To improve regression performance, we propose a novel loss function, the Paired-Siamese Regression loss, which encourages the model to focus on accurately predicting the relative differences between regression label scores rather than just their absolute values. In addition, to gain insight into the brain regions that contribute most strongly to the prediction results, we propose a Critical Region Localization algorithm. This algorithm identifies highly predictive anatomical regions within the white matter fiber tracts for the regression task. We evaluate the effectiveness of the proposed method by predicting individual performance on two neuropsychological assessments of language using a dataset of 20 association white matter fiber tracts from 806 subjects from the Human Connectome Project Young Adult dataset. The results demonstrate superior prediction performance of TractGeoNet compared to several popular regression models that have been applied to predict individual cognitive performance based on neuroimaging features. Of the twenty tracts studied, we find that the left arcuate fasciculus tract is the most highly predictive of the two studied language performance assessments. Within each tract, we localize critical regions whose microstructure and point information are highly and consistently predictive of language performance across different subjects and across multiple independently trained models. These critical regions are widespread and distributed across both hemispheres and all cerebral lobes, including areas of the brain considered important for language function such as superior and anterior temporal regions, pars opercularis, and precentral gyrus. Overall, TractGeoNet demonstrates the potential of geometric deep learning to enhance the study of the brain's white matter fiber tracts and to relate their structure to human traits such as language performance.

Neuromarkers in addiction: definitions, development strategies, and recent advances

Substance use disorders (SUDs) are the most costly and prevalent psychiatric conditions. Recent calls emphasize a need for biomarkers-measurable, stable indicators of normal and abnormal processes and response to treatment or environmental agents-and, in particular, brain-based neuromarkers that will advance understanding of the neurobiological basis of SUDs and clinical practice. To develop neuromarkers, researchers must be grounded in evidence that a putative marker (i) is sensitive and specific to the psychological phenomenon of interest, (ii) constitutes a predictive model, and (iii) generalizes to novel observations (e.g., through internal cross-validation and external application to novel data). These neuromarkers may be used to index risk of developing SUDs (susceptibility), classify individuals with SUDs (diagnostic), assess risk for progression to more severe pathology (prognostic) or index current severity of pathology (monitoring), detect response to treatment (response), and predict individualized treatment outcomes (predictive). Here, we outline guidelines for developing and assessing neuromarkers, we then review recent advances toward neuromarkers in addiction neuroscience centering our discussion around neuromarkers of craving-a core feature of SUDs. In doing so, we specifically focus on the Neurobiological Craving Signature (NCS), which show great promise for meeting the demand of neuromarkers.

Sex classification from functional brain connectivity: Generalization to multiple datasets

Machine learning (ML) approaches are increasingly being applied to neuroimaging data. Studies in neuroscience typically have to rely on a limited set of training data which may impair the generalizability of ML models. However, it is still unclear which kind of training sample is best suited to optimize generalization performance. In the present study, we systematically investigated the generalization performance of sex classification models trained on the parcelwise connectivity profile of either single samples or compound samples of two different sizes. Generalization performance was quantified in terms of mean across-sample classification accuracy and spatial consistency of accurately classifying parcels. Our results indicate that the generalization performance of parcelwise classifiers (pwCs) trained on single dataset samples is dependent on the specific test samples. Certain datasets seem to "match" in the sense that classifiers trained on a sample from one dataset achieved a high accuracy when tested on the respected other one and vice versa. The pwCs trained on the compound samples demonstrated overall highest generalization performance for all test samples, including one derived from a dataset not included in building the training samples. Thus, our results indicate that both a large sample size and a heterogeneous data composition of a training sample have a central role in achieving generalizable results.

Brain decoding of spontaneous thought: Predictive modeling of self-relevance and valence using personal narratives

The contents and dynamics of spontaneous thought are important factors for personality traits and mental health. However, assessing spontaneous thoughts is challenging due to their unconstrained nature, and directing participants' attention to report their thoughts may fundamentally alter them. Here, we aimed to decode two key content dimensions of spontaneous thought-self-relevance and valence-directly from brain activity. To train functional MRI-based predictive models, we used individually generated personal stories as stimuli in a story-reading task to mimic narrative-like spontaneous thoughts (n = 49). We then tested these models on multiple test datasets (total n = 199). The default mode, ventral attention, and frontoparietal networks played key roles in the predictions, with the anterior insula and midcingulate cortex contributing to self-relevance prediction and the left temporoparietal junction and dorsomedial prefrontal cortex contributing to valence prediction. Overall, this study presents brain models of internal thoughts and emotions, highlighting the potential for the brain decoding of spontaneous thought.

Sex classification from functional brain connectivity: Generalization to multiple datasets Generalizability of sex classifiers

Machine learning (ML) approaches are increasingly being applied to neuroimaging data. Studies in neuroscience typically have to rely on a limited set of training data which may impair the generalizability of ML models. However, it is still unclear which kind of training sample is best suited to optimize generalization performance. In the present study, we systematically investigated the generalization performance of sex classification models trained on the parcelwise connectivity profile of either single samples or a compound sample containing data from four different datasets. Generalization performance was quantified in terms of mean across-sample classification accuracy and spatial consistency of accurately classifying parcels. Our results indicate that generalization performance of pwCs trained on single dataset samples is dependent on the specific test samples. Certain datasets seem to "match" in the sense that classifiers trained on a sample from one dataset achieved a high accuracy when tested on the respected other one and vice versa. The pwC trained on the compound sample demonstrated overall highest generalization performance for all test samples, including one derived from a dataset not included in building the training samples. Thus, our results indicate that a big and heterogenous training sample comprising data of multiple datasets is best suited to achieve generalizable results.

Distributed neural representations of conditioned threat in the human brain

Detecting and responding to threat engages several neural nodes including the amygdala, hippocampus, insular cortex, and medial prefrontal cortices. Recent propositions call for the integration of more distributed neural nodes that process sensory and cognitive facets related to threat. Integrative, sensitive, and reproducible distributed neural decoders for the detection and response to threat and safety have yet to be established. We combine functional MRI data across varying threat conditioning and negative affect paradigms from 1465 participants with multivariate pattern analysis to investigate distributed neural representations of threat and safety. The trained decoders sensitively and specifically distinguish between threat and safety cues across multiple datasets. We further show that many neural nodes dynamically shift representations between threat and safety. Our results establish reproducible decoders that integrate neural circuits, merging the well-characterized 'threat circuit' with sensory and cognitive nodes, discriminating threat from safety regardless of experimental designs or data acquisition parameters.

Connectivity-based neuromarker for children's inhibitory control ability and its relevance to body mass index

Preserving a normal body mass index (BMI) is crucial for the healthy growth and development of children. As a core aspect of executive functions, inhibitory control plays a pivotal role in maintaining a normal BMI, which is key to preventing issues of childhood obesity. By studying individual variations in inhibitory control performance and its associated connectivity-based neuromarker in a sample of primary school students (N = 64; 9-12 yr), we aimed to unravel the pathway through which inhibitory control impacts children's BMI. Utilizing resting-state functional MRI scans and a connectivity-based psychometric prediction framework, we found that enhanced inhibitory control abilities were primarily associated with increased functional connectivity in brain structures vital to executive functions, such as the superior frontal lobule, superior parietal lobule, and posterior cingulate cortex. Conversely, inhibitory control abilities displayed a negative relationship with functional connectivity originating from reward-related brain structures, such as the orbital frontal and ventral medial prefrontal lobes. Furthermore, we revealed that both inhibitory control and its corresponding neuromarker can moderate the association between food-related delayed gratification and BMI in children. However, only the neuromarker of inhibitory control maintained its moderating effect on children's future BMI, as determined in the follow-up after one year. Overall, our findings shed light on the potential mechanisms of how inhibitory control in children impacts BMI, highlighting the utility of the connectivity-based neuromarker of inhibitory control in the context of childhood obesity.

Brain-phenotype predictions can survive across diverse real-world data

Recent work suggests that machine learning models predicting psychiatric treatment outcomes based on clinical data may fail when applied to unharmonized samples. Neuroimaging predictive models offer the opportunity to incorporate neurobiological information, which may be more robust to dataset shifts. Yet, among the minority of neuroimaging studies that undertake any form of external validation, there is a notable lack of attention to generalization across dataset-specific idiosyncrasies. Research settings, by design, remove the between-site variations that real-world and, eventually, clinical applications demand. Here, we rigorously test the ability of a range of predictive models to generalize across three diverse, unharmonized samples: the Philadelphia Neurodevelopmental Cohort (n=1291), the Healthy Brain Network (n=1110), and the Human Connectome Project in Development (n=428). These datasets have high inter-dataset heterogeneity, encompassing substantial variations in age distribution, sex, racial and ethnic minority representation, recruitment geography, clinical symptom burdens, fMRI tasks, sequences, and behavioral measures. We demonstrate that reproducible and generalizable brain-behavior associations can be realized across diverse dataset features with sample sizes in the hundreds. Results indicate the potential of functional connectivity-based predictive models to be robust despite substantial inter-dataset variability. Notably, for the HCPD and HBN datasets, the best predictions were not from training and testing in the same dataset (i.e., cross-validation) but across datasets. This result suggests that training on diverse data may improve prediction in specific cases. Overall, this work provides a critical foundation for future work evaluating the generalizability of neuroimaging predictive models in real-world scenarios and clinical settings.

Multi-site benchmark classification of major depressive disorder using machine learning on cortical and subcortical measures

Machine learning (ML) techniques have gained popularity in the neuroimaging field due to their potential for classifying neuropsychiatric disorders. However, the diagnostic predictive power of the existing algorithms has been limited by small sample sizes, lack of representativeness, data leakage, and/or overfitting. Here, we overcome these limitations with the largest multi-site sample size to date (N = 5365) to provide a generalizable ML classification benchmark of major depressive disorder (MDD) using shallow linear and non-linear models. Leveraging brain measures from standardized ENIGMA analysis pipelines in FreeSurfer, we were able to classify MDD versus healthy controls (HC) with a balanced accuracy of around 62%. But after harmonizing the data, e.g., using ComBat, the balanced accuracy dropped to approximately 52%. Accuracy results close to random chance levels were also observed in stratified groups according to age of onset, antidepressant use, number of episodes and sex. Future studies incorporating higher dimensional brain imaging/phenotype features, and/or using more advanced machine and deep learning methods may yield more encouraging prospects.

Revolutionizing tumor detection and classification in multimodality imaging based on deep learning approaches: Methods, applications and limitations

BACKGROUND

The emergence of deep learning (DL) techniques has revolutionized tumor detection and classification in medical imaging, with multimodal medical imaging (MMI) gaining recognition for its precision in diagnosis, treatment, and progression tracking.

OBJECTIVE

This review comprehensively examines DL methods in transforming tumor detection and classification across MMI modalities, aiming to provide insights into advancements, limitations, and key challenges for further progress.

METHODS

Systematic literature analysis identifies DL studies for tumor detection and classification, outlining methodologies including convolutional neural networks (CNNs), recurrent neural networks (RNNs), and their variants. Integration of multimodality imaging enhances accuracy and robustness.

RESULTS

Recent advancements in DL-based MMI evaluation methods are surveyed, focusing on tumor detection and classification tasks. Various DL approaches, including CNNs, YOLO, Siamese Networks, Fusion-Based Models, Attention-Based Models, and Generative Adversarial Networks, are discussed with emphasis on PET-MRI, PET-CT, and SPECT-CT.

FUTURE DIRECTIONS

The review outlines emerging trends and future directions in DL-based tumor analysis, aiming to guide researchers and clinicians toward more effective diagnosis and prognosis. Continued innovation and collaboration are stressed in this rapidly evolving domain.

CONCLUSION

Conclusions drawn from literature analysis underscore the efficacy of DL approaches in tumor detection and classification, highlighting their potential to address challenges in MMI analysis and their implications for clinical practice.

Advancing Naturalistic Affective Science with Deep Learning

People express their own emotions and perceive others' emotions via a variety of channels, including facial movements, body gestures, vocal prosody, and language. Studying these channels of affective behavior offers insight into both the experience and perception of emotion. Prior research has predominantly focused on studying individual channels of affective behavior in isolation using tightly controlled, non-naturalistic experiments. This approach limits our understanding of emotion in more naturalistic contexts where different channels of information tend to interact. Traditional methods struggle to address this limitation: manually annotating behavior is time-consuming, making it infeasible to do at large scale; manually selecting and manipulating stimuli based on hypotheses may neglect unanticipated features, potentially generating biased conclusions; and common linear modeling approaches cannot fully capture the complex, nonlinear, and interactive nature of real-life affective processes. In this methodology review, we describe how deep learning can be applied to address these challenges to advance a more naturalistic affective science. First, we describe current practices in affective research and explain why existing methods face challenges in revealing a more naturalistic understanding of emotion. Second, we introduce deep learning approaches and explain how they can be applied to tackle three main challenges: quantifying naturalistic behaviors, selecting and manipulating naturalistic stimuli, and modeling naturalistic affective processes. Finally, we describe the limitations of these deep learning methods, and how these limitations might be avoided or mitigated. By detailing the promise and the peril of deep learning, this review aims to pave the way for a more naturalistic affective science.

Statistical Learning Methods for Neuroimaging Data Analysis with Applications

The aim of this review is to provide a comprehensive survey of statistical challenges in neuroimaging data analysis, from neuroimaging techniques to large-scale neuroimaging studies and statistical learning methods. We briefly review eight popular neuroimaging techniques and their potential applications in neuroscience research and clinical translation. We delineate four themes of neuroimaging data and review major image processing analysis methods for processing neuroimaging data at the individual level. We briefly review four large-scale neuroimaging-related studies and a consortium on imaging genomics and discuss four themes of neuroimaging data analysis at the population level. We review nine major population-based statistical analysis methods and their associated statistical challenges and present recent progress in statistical methodology to address these challenges.

A Feasibility Study of Diabetic Retinopathy Detection in Type II Diabetic Patients Based on Explainable Artificial Intelligence

Diabetic retinopathy (DR) is vision impairment and a life-threatening condition for diabetic patients. Especially type II diabetic people have higher chances of getting retinal problems. Hence, early prediction of DR is necessary for preventing the diabetic patients from vision impairment. The main aim of this feasibility study is to identify the most critical risk features that could lead to diabetic retinopathy. This study investigated type II diabetic patients' socio-analytical, diabetes, behavioral, and clinical risk factors. We conducted a self-individual questionnaire session for all participants. Our questionnaire asked about the reliability of results, feeling comfortable during the screening test, willingness to participate in future screenings, overall perspective, and satisfaction with the DR screening test. We proposed a random forest model for predicting the prevalence of DR risk among diabetics. Further explanations of the model were conducted using more robust SHAP eXplainable Artificial Intelligence (XAI) tools. The SHAP method makes it possible to understand how input variables interact with their representative output records, as well as how input variables are ranked. In addition, various descriptive and inferential statistical analyses were performed on the data and evaluated the significant relationship between the factors discussed above via hypothesis testing. This feasibility study involved 172 type II diabetic patients (73 males and 99 females). Therefore, we found that 81 (47.09%) out of 172 participants had referable DR. The average age of the patients was determined as 55.08, with a standard deviation of ± 9.770 (ranging from 40 to 79). Type II patients were affected by mild, moderate, severe, and advanced proliferative diabetic retinopathy (PDR) stages with 23.83%, 13.95%, 5.81%, and 3.48%, respectively, of the total samples. The developed RF model obtained high accuracy of 94.9% using clinical dataset. Our results showed that the formation of tiny microminiature lesions was noticeable in type II diabetic patients with aged people, abnormal blood glucose levels, and prolonged diabetes duration.

Subject-specific whole-brain parcellations of nodes and boundaries are modulated differently under 10 Hz rTMS

Repetitive transcranial magnetic stimulation (rTMS) has gained considerable importance in the treatment of neuropsychiatric disorders, including major depression. However, it is not yet understood how rTMS alters brain's functional connectivity. Here we report changes in functional connectivity captured by resting state functional magnetic resonance imaging (rsfMRI) within the first hour after 10 Hz rTMS. We apply subject-specific parcellation schemes to detect changes (1) in network nodes, where the strongest functional connectivity of regions is observed, and (2) in network boundaries, where functional transitions between regions occur. We use support vector machine (SVM), a widely used machine learning algorithm that is robust and effective, for the classification and characterization of time intervals of changes in node and boundary maps. Our results reveal that changes in connectivity at the boundaries are slower and more complex than in those observed in the nodes, but of similar magnitude according to accuracy confidence intervals. These results were strongest in the posterior cingulate cortex and precuneus. As network boundaries are indeed under-investigated in comparison to nodes in connectomics research, our results highlight their contribution to functional adjustments to rTMS.

Music can be reconstructed from human auditory cortex activity using nonlinear decoding models

Music is core to human experience, yet the precise neural dynamics underlying music perception remain unknown. We analyzed a unique intracranial electroencephalography (iEEG) dataset of 29 patients who listened to a Pink Floyd song and applied a stimulus reconstruction approach previously used in the speech domain. We successfully reconstructed a recognizable song from direct neural recordings and quantified the impact of different factors on decoding accuracy. Combining encoding and decoding analyses, we found a right-hemisphere dominance for music perception with a primary role of the superior temporal gyrus (STG), evidenced a new STG subregion tuned to musical rhythm, and defined an anterior-posterior STG organization exhibiting sustained and onset responses to musical elements. Our findings show the feasibility of applying predictive modeling on short datasets acquired in single patients, paving the way for adding musical elements to brain-computer interface (BCI) applications.

A dorsomedial prefrontal cortex-based dynamic functional connectivity model of rumination

Rumination is a cognitive style characterized by repetitive thoughts about one's negative internal states and is a common symptom of depression. Previous studies have linked trait rumination to alterations in the default mode network, but predictive brain markers of rumination are lacking. Here, we adopt a predictive modeling approach to develop a neuroimaging marker of rumination based on the variance of dynamic resting-state functional connectivity and test it across 5 diverse subclinical and clinical samples (total n = 288). A whole-brain marker based on dynamic connectivity with the dorsomedial prefrontal cortex (dmPFC) emerges as generalizable across the subclinical datasets. A refined marker consisting of the most important features from a virtual lesion analysis further predicts depression scores of adults with major depressive disorder (n = 35). This study highlights the role of the dmPFC in trait rumination and provides a dynamic functional connectivity marker for rumination.

Benchmarking explanation methods for mental state decoding with deep learning models

Deep learning (DL) models find increasing application in mental state decoding, where researchers seek to understand the mapping between mental states (e.g., experiencing anger or joy) and brain activity by identifying those spatial and temporal features of brain activity that allow to accurately identify (i.e., decode) these states. Once a DL model has been trained to accurately decode a set of mental states, neuroimaging researchers often make use of methods from explainable artificial intelligence research to understand the model's learned mappings between mental states and brain activity. Here, we benchmark prominent explanation methods in a mental state decoding analysis of multiple functional Magnetic Resonance Imaging (fMRI) datasets. Our findings demonstrate a gradient between two key characteristics of an explanation in mental state decoding, namely, its faithfulness and its alignment with other empirical evidence on the mapping between brain activity and decoded mental state: explanation methods with high explanation faithfulness, which capture the model's decision process well, generally provide explanations that align less well with other empirical evidence than the explanations of methods with less faithfulness. Based on our findings, we provide guidance for neuroimaging researchers on how to choose an explanation method to gain insight into the mental state decoding decisions of DL models.

Machine Learning and Prediction in Fetal, Infant, and Toddler Neuroimaging: A Review and Primer

Predictive models in neuroimaging are increasingly designed with the intent to improve risk stratification and support interventional efforts in psychiatry. Many of these models have been developed in samples of children school-aged or older. Nevertheless, despite growing evidence that altered brain maturation during the fetal, infant, and toddler (FIT) period modulates risk for poor mental health outcomes in childhood, these models are rarely implemented in FIT samples. Applications of predictive modeling in children of these ages provide an opportunity to develop powerful tools for improved characterization of the neural mechanisms underlying development. To facilitate the broader use of predictive models in FIT neuroimaging, we present a brief primer and systematic review on the methods used in current predictive modeling FIT studies. Reflecting on current practices in more than 100 studies conducted over the past decade, we provide an overview of topics, modalities, and methods commonly used in the field and under-researched areas. We then outline ethical and future considerations for neuroimaging researchers interested in predicting health outcomes in early life, including researchers who may be relatively new to either advanced machine learning methods or using FIT data. Altogether, the last decade of FIT research in machine learning has provided a foundation for accelerating the prediction of early-life trajectories across the full spectrum of illness and health.

An Interpretable and Predictive Connectivity-Based Neural Signature for Chronic Cannabis Use

BACKGROUND

Cannabis is one of the most widely used substances in the world, with usage trending upward in recent years. However, although the psychiatric burden associated with maladaptive cannabis use has been well established, reliable and interpretable biomarkers associated with chronic use remain elusive. In this study, we combine large-scale functional magnetic resonance imaging with machine learning and network analysis and develop an interpretable decoding model that offers both accurate prediction and novel insights into chronic cannabis use.

METHODS

Chronic cannabis users (n = 166) and nonusing healthy control subjects (n = 124) completed a cue-elicited craving task during functional magnetic resonance imaging. Linear machine learning methods were used to classify individuals into chronic users and nonusers based on whole-brain functional connectivity. Network analysis was used to identify the most predictive regions and communities.

RESULTS

We obtained high (∼80% out-of-sample) accuracy across 4 different classification models, demonstrating that task-evoked connectivity can successfully differentiate chronic cannabis users from nonusers. We also identified key predictive regions implicating motor, sensory, attention, and craving-related areas, as well as a core set of brain networks that contributed to successful classification. The most predictive networks also strongly correlated with cannabis craving within the chronic user group.

CONCLUSIONS

This novel approach produced a neural signature of chronic cannabis use that is both accurate in terms of out-of-sample prediction and interpretable in terms of predictive networks and their relation to cannabis craving.

Subgroups of internet gaming disorder based on addiction-related resting-state functional connectivity

AIMS

To identify subgroups of people with internet gaming disorder (IGD) based on addiction-related resting-state functional connectivity and how these subgroups show different clinical correlates and responses to treatment.

DESIGN

Secondary analysis of two functional magnetic resonance imaging (fMRI) data sets.

SETTING

Zhejiang province and Beijing, China.

PARTICIPANTS

One hundred and sixty-nine IGD and 147 control subjects.

MEASUREMENTS

k-Means algorithmic and support-vector machine-learning approaches were used to identify subgroups of IGD subjects. These groups were examined with respect to assessments of craving, behavioral activation and inhibition, emotional regulation, cue-reactivity and guessing-related measures.

FINDINGS

Two groups of subjects with IGD were identified and defined by distinct patterns of connectivity in brain networks previously implicated in addictions: subgroup 1 ('craving-related subgroup') and subgroup 2 ('mixed psychological subgroup'). Clustering IGD on this basis enabled the development of diagnostic classifiers with high sensitivity and specificity for IGD subgroups in 10-fold validation (n = 218) and out-of-sample replication (n = 98) data sets. Subgroup 1 is characterized by high craving scores, cue-reactivity during fMRI and responsiveness to a craving behavioral intervention therapy. Subgroup 2 is characterized by high craving, behavioral inhibition and activations scores, non-adaptive emotion-regulation strategies and guessing-task fMRI measures. Subgroups 1 and 2 showed largely opposite functional-connectivity patterns in overlapping networks.

CONCLUSIONS

There appear to be two subgroups of people with internet gaming disorder, each associated with differing patterns of brain functional connectivity and distinct clinical symptom profiles and gender compositions.

Confounds in neuroimaging: A clear case of sex as a confound in brain-based prediction

Muscle weakness is common in many neurological, neuromuscular, and musculoskeletal conditions. Muscle size only partially explains muscle strength as adaptions within the nervous system also contribute to strength. Brain-based biomarkers of neuromuscular function could provide diagnostic, prognostic, and predictive value in treating these disorders. Therefore, we sought to characterize and quantify the brain's contribution to strength by developing multimodal MRI pipelines to predict grip strength. However, the prediction of strength was not straightforward, and we present a case of sex being a clear confound in brain decoding analyses. While each MRI modality-structural MRI (i.e., gray matter morphometry), diffusion MRI (i.e., white matter fractional anisotropy), resting state functional MRI (i.e., functional connectivity), and task-evoked functional MRI (i.e., left or right hand motor task activation)-and a multimodal prediction pipeline demonstrated significant predictive power for strength (R ² = 0.108-0.536, p ≤ 0.001), after correcting for sex, the predictive power was substantially reduced (R ² = -0.038-0.075). Next, we flipped the analysis and demonstrated that each MRI modality and a multimodal prediction pipeline could significantly predict sex (accuracy = 68.0%-93.3%, AUC = 0.780-0.982, p < 0.001). However, correcting the brain features for strength reduced the accuracy for predicting sex (accuracy = 57.3%-69.3%, AUC = 0.615-0.780). Here we demonstrate the effects of sex-correlated confounds in brain-based predictive models across multiple brain MRI modalities for both regression and classification models. We discuss implications of confounds in predictive modeling and the development of brain-based MRI biomarkers, as well as possible strategies to overcome these barriers.

Explainable AI: A review of applications to neuroimaging data

Deep neural networks (DNNs) have transformed the field of computer vision and currently constitute some of the best models for representations learned via hierarchical processing in the human brain. In medical imaging, these models have shown human-level performance and even higher in the early diagnosis of a wide range of diseases. However, the goal is often not only to accurately predict group membership or diagnose but also to provide explanations that support the model decision in a context that a human can readily interpret. The limited transparency has hindered the adoption of DNN algorithms across many domains. Numerous explainable artificial intelligence (XAI) techniques have been developed to peer inside the "black box" and make sense of DNN models, taking somewhat divergent approaches. Here, we suggest that these methods may be considered in light of the interpretation goal, including functional or mechanistic interpretations, developing archetypal class instances, or assessing the relevance of certain features or mappings on a trained model in a post-hoc capacity. We then focus on reviewing recent applications of post-hoc relevance techniques as applied to neuroimaging data. Moreover, this article suggests a method for comparing the reliability of XAI methods, especially in deep neural networks, along with their advantages and pitfalls.

Identifying and distinguishing of essential tremor and Parkinson's disease with grouped stability analysis based on searchlight-based MVPA

BACKGROUND

Since both essential tremor (ET) and Parkinson's disease (PD) are movement disorders and share similar clinical symptoms, it is very difficult to recognize the differences in the presentation, course, and treatment of ET and PD, which leads to misdiagnosed commonly.

PURPOSE

Although neuroimaging biomarker of ET and PD has been investigated based on statistical analysis, it is unable to assist the clinical diagnosis of ET and PD and ensure the efficiency of these biomarkers. The aim of the study was to identify the neuroimaging biomarkers of ET and PD based on structural magnetic resonance imaging (MRI). Moreover, the study also distinguished ET from PD via these biomarkers to validate their classification performance.

METHODS

This study has developed and implemented a three-level machine learning framework to identify and distinguish ET and PD. First of all, at the model-level assessment, the searchlight-based machine learning method has been used to identify the group differences of patients (ET/PD) with normal controls (NCs). And then, at the feature-level assessment, the stability of group differences has been tested based on structural brain atlas separately using the permutation test to identify the robust neuroimaging biomarkers. Furthermore, the identified biomarkers of ET and PD have been applied to classify ET from PD based on machine learning techniques. Finally, the identified biomarkers have been compared with the previous findings of the biology-level assessment.

RESULTS

According to the biomarkers identified by machine learning, this study has found widespread alterations of gray matter (GM) for ET and large overlap between ET and PD and achieved superior classification performance (PCA + SVM, accuracy = 100%).

CONCLUSIONS

This study has demonstrated the significance of a machine learning framework to identify and distinguish ET and PD. Future studies using a large data set are needed to confirm the potential clinical application of machine learning techniques to discern between PD and ET.

Interpreting mental state decoding with deep learning models

In mental state decoding, researchers aim to identify the set of mental states (e.g., experiencing happiness or fear) that can be reliably identified from the activity patterns of a brain region (or network). Deep learning (DL) models are highly promising for mental state decoding because of their unmatched ability to learn versatile representations of complex data. However, their widespread application in mental state decoding is hindered by their lack of interpretability, difficulties in applying them to small datasets, and in ensuring their reproducibility and robustness. We recommend approaching these challenges by leveraging recent advances in explainable artificial intelligence (XAI) and transfer learning, and also provide recommendations on how to improve the reproducibility and robustness of DL models in mental state decoding.

Proportional intracranial volume correction differentially biases behavioral predictions across neuroanatomical features, sexes, and development

Individual differences in brain anatomy can be used to predict variations in cognitive ability. Most studies to date have focused on broad population-level trends, but the extent to which the observed predictive features are shared across sexes and age groups remains to be established. While it is standard practice to account for intracranial volume (ICV) using proportion correction in both regional and whole-brain morphometric analyses, in the context of brain-behavior predictions the possible differential impact of ICV correction on anatomical features and subgroups within the population has yet to be systematically investigated. In this work, we evaluate the effect of proportional ICV correction on sex-independent and sex-specific predictive models of individual cognitive abilities across multiple anatomical properties (surface area, gray matter volume, and cortical thickness) in healthy young adults (Human Connectome Project; n = 1013, 548 females) and typically developing children (Adolescent Brain Cognitive Development study; n = 1823, 979 females). We demonstrate that ICV correction generally reduces predictive accuracies derived from surface area and gray matter volume, while increasing predictive accuracies based on cortical thickness in both adults and children. Furthermore, the extent to which predictive models generalize across sexes and age groups depends on ICV correction: models based on surface area and gray matter volume are more generalizable without ICV correction, while models based on cortical thickness are more generalizable with ICV correction. Finally, the observed neuroanatomical features predictive of cognitive abilities are unique across age groups regardless of ICV correction, but whether they are shared or unique across sexes (within age groups) depends on ICV correction. These findings highlight the importance of considering individual differences in ICV, and show that proportional ICV correction does not remove the effects of cranial volume from anatomical measurements and can introduce ICV bias where previously there was none. ICV correction choices affect not just the strength of the relationships captured, but also the conclusions drawn regarding the neuroanatomical features that underlie those relationships.

A hybrid learning framework for fine-grained interpretation of brain spatiotemporal patterns during naturalistic functional magnetic resonance imaging

Naturalistic stimuli, including movie, music, and speech, have been increasingly applied in the research of neuroimaging. Relative to a resting-state or single-task state, naturalistic stimuli can evoke more intense brain activities and have been proved to possess higher test–retest reliability, suggesting greater potential to study adaptive human brain function. In the current research, naturalistic functional magnetic resonance imaging (N-fMRI) has been a powerful tool to record brain states under naturalistic stimuli, and many efforts have been devoted to study the high-level semantic features from spatial or temporal representations via N-fMRI. However, integrating both spatial and temporal characteristics of brain activities for better interpreting the patterns under naturalistic stimuli is still underexplored. In this work, a novel hybrid learning framework that comprehensively investigates both the spatial (via Predictive Model) and the temporal [via convolutional neural network (CNN) model] characteristics of the brain is proposed. Specifically, to focus on certain relevant regions from the whole brain, regions of significance (ROS), which contain common spatial activation characteristics across individuals, are selected via the Predictive Model. Further, voxels of significance (VOS), whose signals contain significant temporal characteristics under naturalistic stimuli, are interpreted via one-dimensional CNN (1D-CNN) model. In this article, our proposed framework is applied onto the N-fMRI data during naturalistic classical/pop/speech audios stimuli. The promising performance is achieved via the Predictive Model to differentiate the different audio categories. Especially for distinguishing the classic and speech audios, the accuracy of classification is up to 92%. Moreover, spatial ROS and VOS are effectively obtained. Besides, temporal characteristics of the high-level semantic features are investigated on the frequency domain via convolution kernels of 1D-CNN model, and we effectively bridge the “semantic gap” between high-level semantic features of N-fMRI and low-level acoustic features of naturalistic audios in the frequency domain. Our results provide novel insights on characterizing spatiotemporal patterns of brain activities via N-fMRI and effectively explore the high-level semantic features under naturalistic stimuli, which will further benefit the understanding of the brain working mechanism and the advance of naturalistic stimuli clinical application.

Demographic reporting across a decade of neuroimaging: a systematic review

Diversity of participants in biomedical research with respect to race, ethnicity, and biological sex is crucial, particularly given differences in disease prevalence, recovery, and survival rates between demographic groups. The objective of this systematic review was to report on the demographics of neuroimaging studies using magnetic resonance imaging (MRI). The Web of Science database was used and data collection was performed between June 2021 to November 2021; all articles were reviewed independently by at least two researchers. Articles utilizing MR data acquired in the United States, with n ≥ 10 human subjects, and published between 2010–2020 were included. Non-primary research articles and those published in journals that did not meet a quality control check were excluded. Of the 408 studies meeting inclusion criteria, approximately 77% report sex, 10% report race, and 4% report ethnicity. Demographic reporting also varied as function of disease studied, participant age range, funding, and publisher. We anticipate quantitative data on the extent, or lack, of reporting will be necessary to ensure inclusion of diverse populations in biomedical research.

Choice of Voxel-based Morphometry processing pipeline drives variability in the location of neuroanatomical brain markers

Fundamental and clinical neuroscience has benefited tremendously from the development of automated computational analyses. In excess of 600 human neuroimaging papers using Voxel-based Morphometry (VBM) are now published every year and a number of different automated processing pipelines are used, although it remains to be systematically assessed whether they come up with the same answers. Here we examined variability between four commonly used VBM pipelines in two large brain structural datasets. Spatial similarity and between-pipeline reproducibility of the processed gray matter brain maps were generally low between pipelines. Examination of sex-differences and age-related changes revealed considerable differences between the pipelines in terms of the specific regions identified. Machine learning-based multivariate analyses allowed accurate predictions of sex and age, however accuracy differed between pipelines. Our findings suggest that the choice of pipeline alone leads to considerable variability in brain structural markers which poses a serious challenge for reproducibility and interpretation.

Four common processing pipelines tested on two Voxel-based Morphometry (VBM) datasets yield considerable variations in results, raising issues on the interpretability and robustness of VBM results.

When self comes to a wandering mind: Brain representations and dynamics of self-generated concepts in spontaneous thought

Self-relevant concepts are major building blocks of spontaneous thought, and their dynamics in a natural stream of thought are likely to reveal one's internal states that are important for mental health. Here, we conducted a functional magnetic resonance imaging experiment (n = 62) to examine brain representations and dynamics of self-generated concepts in the context of spontaneous thought using a newly developed free association-based thought sampling task. The dynamics of conceptual associations were predictive of individual differences in general negative affectivity, replicating across multiple datasets (n = 196). Reflecting on self-generated concepts strongly engaged brain regions linked to autobiographical memory, conceptual processes, emotion, and autonomic regulation, including the medial prefrontal and medial temporal subcortical structures. Multivariate pattern-based predictive modeling revealed that the neural representations of valence became more person-specific as the level of perceived self-relevance increased. Overall, this study sheds light on how self-generated concepts in spontaneous thought construct inner affective states and idiosyncrasies.

The expanding horizons of network neuroscience: From description to prediction and control

The field of network neuroscience has emerged as a natural framework for the study of the brain and has been increasingly applied across divergent problems in neuroscience. From a disciplinary perspective, network neuroscience originally emerged as a formal integration of graph theory (from mathematics) and neuroscience (from biology). This early integration afforded marked utility in describing the interconnected nature of neural units, both structurally and functionally, and underscored the relevance of that interconnection for cognition and behavior. But since its inception, the field has not remained static in its methodological composition. Instead, it has grown to use increasingly advanced graph-theoretic tools and to bring in several other disciplinary perspectives-including machine learning and systems engineering-that have proven complementary. In doing so, the problem space amenable to the discipline has expanded markedly. In this review, we discuss three distinct flavors of investigation in state-of-the-art network neuroscience: (i) descriptive network neuroscience, (ii) predictive network neuroscience, and (iii) a perturbative network neuroscience that draws on recent advances in network control theory. In considering each area, we provide a brief summary of the approaches, discuss the nature of the insights obtained, and highlight future directions.

Gender-related differences in involvement of addiction brain networks in internet gaming disorder: Relationships with craving and emotional regulation

BACKGROUND

Abnormal interactions among addiction brain networks associated with intoxication, negative affect, and anticipation may have relevance for internet gaming disorder (IGD). Despite prior studies having identified gender-related differences in the neural correlates of IGD, gender-related differences in the involvement of brain networks remain unclear.

METHODS

One-hundred-and-nine individuals with IGD (54 males) and 111 with recreational game use (RGU; 58 males) provided resting-state fMRI data. We examined gender-related differences in involvement of addiction brain networks in IGD versus RGU subjects. We further compared the strength between and within addiction brain networks and explored possible relationships between the strength of functional connectivities within and between addiction brain networks and several relevant behavioral measures.

RESULTS

The addiction brain networks showed high correct classification rates in distinguishing IGD and RGU subjects in men and women. Male subjects with versus without IGD showed stronger functional connectivities between and within addiction brain networks. Moreover, the strength of the connectivity within the anticipation network in male IGD subjects was positively related to subjective craving. However, female subjects with versus without IGD showed decreased functional connections between and within addiction brain networks. The strength of connectivity between the anticipation and negative-affect brain networks in female IGD subjects was negatively related to maladaptive cognitive emotion-regulation strategies.

CONCLUSIONS

Addiction brain networks have potential for distinguishing IGD and RGU individuals. Importantly, this study identified novel gender-related differences in brain-behavior relationships in IGD. These results help advance current neuroscientific theories of IGD and may inform gender-informed treatment strategies.

Linking cerebellar functional gradients to transdiagnostic behavioral dimensions of psychopathology

High co-morbidity and substantial overlap across psychiatric disorders encourage a transition in psychiatry research from categorical to dimensional approaches that integrate neuroscience and psychopathology. Converging evidence suggests that the cerebellum is involved in a wide range of cognitive functions and mental disorders. An important question thus centers on the extent to which cerebellar function can be linked to transdiagnostic dimensions of psychopathology. To address this question, we used a multivariate data-driven statistical technique (partial least squares) to identify latent dimensions linking human cerebellar connectome as assessed by functional MRI to a large set of clinical, cognitive, and trait measures across 198 participants, including healthy controls (n = 92) as well as patients diagnosed with attention-deficit/hyperactivity disorder (n = 35), bipolar disorder (n = 36), and schizophrenia (n = 35). Macroscale spatial gradients of connectivity at voxel level were used to characterize cerebellar connectome properties, which provide a low-dimensional representation of cerebellar connectivity, i.e., a sensorimotor-supramodal hierarchical organization. This multivariate analysis revealed significant correlated patterns of cerebellar connectivity gradients and behavioral measures that could be represented into four latent dimensions: general psychopathology, impulsivity and mood, internalizing symptoms and executive dysfunction. Each dimension was associated with a unique spatial pattern of cerebellar connectivity gradients across all participants. Multiple control analyses and 10-fold cross-validation confirmed the robustness and generalizability of the yielded four dimensions. These findings highlight the relevance of cerebellar connectivity as a necessity for the study and classification of transdiagnostic dimensions of psychopathology and call on researcher to pay more attention to the role of cerebellum in the dimensions of psychopathology, not just within the cerebral cortex.

Intrinsic brain activity patterns across large-scale networks predict reciprocity propensity

Reciprocity is prevalent across human societies, but individuals are heterogeneous regarding their reciprocity propensity. Although a large body of task-based brain imaging measures has shed light on the neural underpinnings of reciprocity at group level, the neural basis underlying the individual differences in reciprocity propensity remains largely unclear. Here, we combined brain imaging and machine learning techniques to individually predict reciprocity propensity from resting-state brain activity measured by fractional amplitude of low-frequency fluctuation. The brain regions contributing to the prediction were then analyzed for functional connectivity and decoding analyses, allowing for a data-driven quantitative inference on psychophysiological functions. Our results indicated that patterns of resting-state brain activity across multiple brain systems were capable of predicting individual reciprocity propensity, with the contributing regions distributed across the salience (e.g., ventrolateral prefrontal cortex), fronto-parietal (e.g., dorsolateral prefrontal cortex), default mode (e.g., ventromedial prefrontal cortex), and sensorimotor (e.g., supplementary motor area) networks. Those contributing brain networks are implicated in emotion and cognitive control, mentalizing, and motor-based processes, respectively. Collectively, these findings provide novel evidence on the neural signatures underlying the individual differences in reciprocity, and lend support the assertion that reciprocity emerges from interactions among regions embodied in multiple large-scale brain networks.

A Neuroimaging Signature of Cognitive Aging from Whole-Brain Functional Connectivity

Cognitive decline is amongst one of the most commonly reported complaints during normal aging. Despite evidence that age and cognition are linked with similar neural correlates, no previous studies have directly ascertained how these two constructs overlap in the brain in terms of neuroimaging-based prediction. Based on a long lifespan healthy cohort (CamCAN, aged 19-89 years, n = 567), it is shown that both cognitive function (domains spanning executive function, emotion processing, motor function, and memory) and human age can be reliably predicted from unique patterns of functional connectivity, with models generalizable in two external datasets (n = 533 and n = 453). Results show that cognitive decline and normal aging both manifest decrease within-network connections (especially default mode and ventral attention networks) and increase between-network connections (somatomotor network). Whereas dorsal attention network is an exception, which is highly predictive on cognitive ability but is weakly correlated with aging. Further, the positively weighted connections in predicting fluid intelligence significantly mediate its association with age. Together, these findings offer insights into why normal aging is often associated with cognitive decline in terms of brain network organization, indicating a process of neural dedifferentiation and compensational theory.

Predicting the future of neuroimaging predictive models in mental health

Predictive modeling using neuroimaging data has the potential to improve our understanding of the neurobiology underlying psychiatric disorders and putatively information interventions. Accordingly, there is a plethora of literature reviewing published studies, the mathematics underlying machine learning, and the best practices for using these approaches. As our knowledge of mental health and machine learning continue to evolve, we instead aim to look forward and "predict" topics that we believe will be important in current and future studies. Some of the most discussed topics in machine learning, such as bias and fairness, the handling of dirty data, and interpretable models, may be less familiar to the broader community using neuroimaging-based predictive modeling in psychiatry. In a similar vein, transdiagnostic research and targeting brain-based features for psychiatric intervention are modern topics in psychiatry that predictive models are well-suited to tackle. In this work, we target an audience who is a researcher familiar with the fundamental procedures of machine learning and who wishes to increase their knowledge of ongoing topics in the field. We aim to accelerate the utility and applications of neuroimaging-based predictive models for psychiatric research by highlighting and considering these topics. Furthermore, though not a focus, these ideas generalize to neuroimaging-based predictive modeling in other clinical neurosciences and predictive modeling with different data types (e.g., digital health data).