Predicting visual working memory with multimodal magnetic resonance imaging

Multimodal covarying brain patterns mediate genetic and psychological contributions to individual differences in pain sensitivity

ABSTRACT

Individuals vary significantly in their pain sensitivity, with contributions from the brain, genes, and psychological factors. However, a multidimensional model integrating these factors is lacking due to their complex interactions. To address this, we measured pain sensitivity (ie, pain threshold and pain tolerance) using the cold pressor test, collected magnetic resonance imaging (MRI) data and genetic data, and evaluated psychological factors (ie, pain catastrophizing, pain-related fear, and pain-related anxiety) from 450 healthy participants with both sexes (160 male, 290 female). Using multimodal MRI fusion methods, we identified 2 pairs of covarying structural and functional brain patterns associated with pain threshold and tolerance, respectively. These patterns primarily involved regions related to self-awareness, sensory-discriminative, cognitive-evaluative, motion preparation and execution, and emotional aspects of pain. Notably, pain catastrophizing was negatively correlated with pain tolerance, and this relationship was mediated by the multimodal covarying brain patterns in male participants only. Furthermore, we identified an association between the single-nucleotide polymorphism rs4141964 within the fatty acid amide hydrolase gene and pain threshold, mediated by the identified multimodal covarying brain patterns across all participants. In summary, we suggested a model that integrates the brain, genes, and psychological factors to elucidate their role in shaping interindividual variations in pain sensitivity, highlighting the important contribution of the multimodal covarying brain patterns as important biological mediators in the associations between genes/psychological factors and pain sensitivity.

Prediction of cognitive performance differences in older age from multimodal neuroimaging data

Differences in brain structure and functional and structural network architecture have been found to partly explain cognitive performance differences in older ages. Thus, they may serve as potential markers for these differences. Initial unimodal studies, however, have reported mixed prediction results of selective cognitive variables based on these brain features using machine learning (ML). Thus, the aim of the current study was to investigate the general validity of cognitive performance prediction from imaging data in healthy older adults. In particular, the focus was with examining whether (1) multimodal information, i.e., region-wise grey matter volume (GMV), resting-state functional connectivity (RSFC), and structural connectivity (SC) estimates, may improve predictability of cognitive targets, (2) predictability differences arise for global cognition and distinct cognitive profiles, and (3) results generalize across different ML approaches in 594 healthy older adults (age range: 55-85 years) from the 1000BRAINS study. Prediction potential was examined for each modality and all multimodal combinations, with and without confound (i.e., age, education, and sex) regression across different analytic options, i.e., variations in algorithms, feature sets, and multimodal approaches (i.e., concatenation vs. stacking). Results showed that prediction performance differed considerably between deconfounding strategies. In the absence of demographic confounder control, successful prediction of cognitive performance could be observed across analytic choices. Combination of different modalities tended to marginally improve predictability of cognitive performance compared to single modalities. Importantly, all previously described effects vanished in the strict confounder control condition. Despite a small trend for a multimodal benefit, developing a biomarker for cognitive aging remains challenging.

Blood pressure-related white matter microstructural disintegrity and associated cognitive function impairment in asymptomatic adults

BACKGROUND AND OBJECTIVES

We aimed to investigate the white matter (WM) microstructural/cytostructural disintegrity patterns related to higher systolic blood pressure (SBP), and whether they mediate SBP effects on cognitive performance in middle-aged adults.

METHODS

Using the UK Biobank study of community-dwelling volunteers aged 40-69 years, we included participants without a history of stroke, dementia, demyelinating disease or traumatic brain injury. We investigated the association of SBP with MRI diffusion metrics: fractional anisotropy (FA), mean diffusivity (MD), intracellular volume fraction (a measure of neurite density), isotropic (free) water volume fraction (ISOVF) and orientation dispersion across WM tracts. Then, we determined whether WM diffusion metrics mediated the effects of SBP on cognitive function.

RESULTS

We analysed 31 363 participants-mean age of 63.8 years (SD: 7.7), and 16 523 (53%) females. Higher SBP was associated with lower FA and neurite density, but higher MD and ISOVF. Among different WM tracts, diffusion metrics of the internal capsule anterior limb, external capsule, superior and posterior corona radiata were most affected by higher SBP. Among seven cognitive metrics, SBP levels were only associated with 'fluid intelligence' (adjusted p<0.001). In mediation analysis, the averaged FA of external capsule, internal capsule anterior limb and superior cerebellar peduncle mediated 13%, 9% and 13% of SBP effects on fluid intelligence, while the averaged MD of external capsule, internal capsule anterior and posterior limbs, and superior corona radiata mediated 5%, 7%, 7% and 6% of SBP effects on fluid intelligence, respectively.

DISCUSSION

Among asymptomatic adults, higher SBP is associated with pervasive WM microstructure disintegrity, partially due to reduced neuronal count, which appears to mediate SBP adverse effects on fluid intelligence. Diffusion metrics of select WM tracts, which are most reflective of SBP-related parenchymal damage and cognitive impairment, may serve as imaging biomarkers to assess treatment response in antihypertensive trials.

The challenges and prospects of brain-based prediction of behaviour

Relating individual brain patterns to behaviour is fundamental in system neuroscience. Recently, the predictive modelling approach has become increasingly popular, largely due to the recent availability of large open datasets and access to computational resources. This means that we can use machine learning models and interindividual differences at the brain level represented by neuroimaging features to predict interindividual differences in behavioural measures. By doing so, we could identify biomarkers and neural correlates in a data-driven fashion. Nevertheless, this budding field of neuroimaging-based predictive modelling is facing issues that may limit its potential applications. Here we review these existing challenges, as well as those that we anticipate as the field develops. We focus on the impacts of these challenges on brain-based predictions. We suggest potential solutions to address the resolvable challenges, while keeping in mind that some general and conceptual limitations may also underlie the predictive modelling approach.

Relationship Between Prediction Accuracy and Feature Importance Reliability: an Empirical and Theoretical Study

There is significant interest in using neuroimaging data to predict behavior. The predictive models are often interpreted by the computation of feature importance, which quantifies the predictive relevance of an imaging feature. Tian and Zalesky (2021) suggest that feature importance estimates exhibit low split-half reliability, as well as a trade-off between prediction accuracy and feature importance reliability across parcellation resolutions. However, it is unclear whether the trade-off between prediction accuracy and feature importance reliability is universal. Here, we demonstrate that, with a sufficient sample size, feature importance (operationalized as Haufe-transformed weights) can achieve fair to excellent split-half reliability. With a sample size of 2600 participants, Haufe-transformed weights achieve average intra-class correlation coefficients of 0.75, 0.57 and 0.53 for cognitive, personality and mental health measures respectively. Haufe-transformed weights are much more reliable than original regression weights and univariate FC-behavior correlations. Original regression weights are not reliable even with 2600 participants. Intriguingly, feature importance reliability is strongly positively correlated with prediction accuracy across phenotypes. Within a particular behavioral domain, there is no clear relationship between prediction performance and feature importance reliability across regression models. Furthermore, we show mathematically that feature importance reliability is necessary, but not sufficient, for low feature importance error. In the case of linear models, lower feature importance error is mathematically related to lower prediction error. Therefore, higher feature importance reliability might yield lower feature importance error and higher prediction accuracy. Finally, we discuss how our theoretical results relate with the reliability of imaging features and behavioral measures. Overall, the current study provides empirical and theoretical insights into the relationship between prediction accuracy and feature importance reliability.

Comparison of individualized behavioral predictions across anatomical, diffusion and functional connectivity MRI

A fundamental goal across the neurosciences is the characterization of relationships linking brain anatomy, functioning, and behavior. Although various MRI modalities have been developed to probe these relationships, direct comparisons of their ability to predict behavior have been lacking. Here, we compared the ability of anatomical T1, diffusion and functional MRI (fMRI) to predict behavior at an individual level. Cortical thickness, area and volume were extracted from anatomical T1 images. Diffusion Tensor Imaging (DTI) and approximate Neurite Orientation Dispersion and Density Imaging (NODDI) models were fitted to the diffusion images. The resulting metrics were projected to the Tract-Based Spatial Statistics (TBSS) skeleton. We also ran probabilistic tractography for the diffusion images, from which we extracted the stream count, average stream length, and the average of each DTI and NODDI metric across tracts connecting each pair of brain regions. Functional connectivity (FC) was extracted from both task and resting-state fMRI. Individualized prediction of a wide range of behavioral measures were performed using kernel ridge regression, linear ridge regression and elastic net regression. Consistency of the results were investigated with the Human Connectome Project (HCP) and Adolescent Brain Cognitive Development (ABCD) datasets. In both datasets, FC-based models gave the best prediction performance, regardless of regression model or behavioral measure. This was especially true for the cognitive component. Furthermore, all modalities were able to predict cognition better than other behavioral components. Combining all modalities improved prediction of cognition, but not other behavioral components. Finally, across all behaviors, combining resting and task FC yielded prediction performance similar to combining all modalities. Overall, our study suggests that in the case of healthy children and young adults, behaviorally-relevant information in T1 and diffusion features might reflect a subset of the variance captured by FC.

Functional connectivity dynamics as a function of the fluctuation of tension during film watching

To advance the understanding of the dynamic relationship between brain activities and emotional experiences, we examined the neural patterns of tension, a unique emotion that highly depends on how an event unfolds. Specifically, the present study explored the temporal relationship between functional connectivity patterns within and between different brain functional modules and the fluctuation in tension during film watching. Due to the highly contextualized and time-varying nature of tension, we expected that multiple neural networks would be involved in the dynamic tension experience. Using the neuroimaging data of 546 participants, we conducted a dynamic brain analysis to identify the intra- and inter-module functional connectivity patterns that are significantly correlated with the fluctuation of tension over time. The results showed that the inter-module connectivity of cingulo-opercular network, fronto-parietal network, and default mode network is involved in the dynamic experience of tension. These findings demonstrate a close relationship between brain functional connectivity patterns and emotional dynamics, which supports the importance of functional connectivity dynamics in understanding our cognitive and emotional processes.

Combining Regional and Connectivity Metrics of Functional Magnetic Resonance Imaging and Diffusion Tensor Imaging for Individualized Prediction of Pain Sensitivity

Characterization and prediction of individual difference of pain sensitivity are of great importance in clinical practice. MRI techniques, such as functional magnetic resonance imaging (fMRI) and diffusion tensor imaging (DTI), have been popularly used to predict an individual's pain sensitivity, but existing studies are limited by using one single imaging modality (fMRI or DTI) and/or using one type of metrics (regional or connectivity features). As a result, pain-relevant information in MRI has not been fully revealed and the associations among different imaging modalities and different features have not been fully explored for elucidating pain sensitivity. In this study, we investigated the predictive capability of multi-features (regional and connectivity metrics) of multimodal MRI (fMRI and DTI) in the prediction of pain sensitivity using data from 210 healthy subjects. We found that fusing fMRI-DTI and regional-connectivity features are capable of more accurately predicting an individual's pain sensitivity than only using one type of feature or using one imaging modality. These results revealed rich information regarding individual pain sensitivity from the brain's both structural and functional perspectives as well as from both regional and connectivity metrics. Hence, this study provided a more comprehensive characterization of the neural correlates of individual pain sensitivity, which holds a great potential for clinical pain management.

Widespread White Matter Microstructure Alterations Based on Diffusion Tensor Imaging and Diffusion Kurtosis Imaging in Patients With Pontine Infarction

Neurological deficits after stroke are closely related to white matter microstructure damage. However, secondary changes in white matter microstructure after pontine infarction (PI) in the whole brain remain unclear. This study aimed to investigate the correlation of diffusion kurtosis imaging (DKI)-derived diffusion and kurtosis parameters of abnormal white matter tracts with behavioral function in patients with chronic PI. Overall, 60 patients with unilateral chronic PI (33 patients with left PI and 27 patients with right PI) and 30 normal subjects were recruited and underwent DKI scans. Diffusion parameters derived from diffusion tensor imaging (DTI) and DKI and kurtosis parameters derived from DKI were obtained. Between-group differences in multiple parameters were analyzed to assess the changes in abnormal white matter microstructure. Moreover, we also calculated the sensitivities of different diffusion and kurtosis parameters of DTI and DKI for identifying abnormal white matter tracts. Correlations between the DKI-derived parameters in secondary microstructure changes and behavioral scores in the PI were analyzed. Compared with the NC group, both left PI and right PI groups showed more extensive perilesional and remote white matter microstructure changes. The DKI-derived diffusion parameters showed higher sensitivities than did the DTI-derived parameters. Further, DKI-derived diffusion and kurtosis parameters in abnormal white matter regions were correlated with impaired motor and cognitive function in patients with PI. In conclusion, PI could lead to extensive white matter tracts impairment in perilesional and remote regions. Further, the diffusion and kurtosis parameters could be complementary for identifying comprehensive tissue microstructural damage after PI.

Identification of violent patients with schizophrenia using a hybrid machine learning approach at the individual level

Despite numerous risk factors associated with violence in patients with schizophrenia, predicting and preventing violent behavior is still a challenge. At present, machine learning (ML) has become a promising strategy for guiding individualized assessment. To build an effective model to predict the risk of violence in patients with schizophrenia, we proposed a hybrid ML method to improve the prediction capability in 42 violent offenders with schizophrenia and 33 non-violent patients with schizophrenia. The results revealed that the final model, which combined multimodal data, achieved the highest prediction performance with an accuracy of 90.67%. Specifically, the model, which fused three modalities of neuroimaging data, achieved a better accuracy than other fused models. In addition, the msot discriminative neuroimaging features involved in the prefrontal-temporal cognitive circuit and striatum reward system, indicating that dysfunction in cortical-subcortical circuits might be associated with high risk of violence in patients with schizophrenia. This study provides the first evidence supporting that the combination of specific multimodal neuroimaging and clinical data in ML analysis can effectively identify violent patients with schizophrenia. Furthermore, this work is crucial for the development of neuro-prediction models that could facilitate individualized treatment and interventions for violent behaviors in patients with schizophrenia.

Predicting visual working memory with multimodal magnetic resonance imaging

The indispensability of visual working memory (VWM) in human daily life suggests its importance in higher cognitive functions and neurological diseases. However, despite the extensive research efforts, most findings on the neural basis of VWM are limited to a unimodal context (either structure or function) and have low generalization. To address the above issues, this study proposed the usage of multimodal neuroimaging in combination with machine learning to reveal the neural mechanism of VWM across a large cohort (N = 547). Specifically, multimodal magnetic resonance imaging features extracted from voxel-wise amplitude of low-frequency fluctuations, gray matter volume, and fractional anisotropy were used to build an individual VWM capacity prediction model through a machine learning pipeline, including the steps of feature selection, relevance vector regression, cross-validation, and model fusion. The resulting model exhibited promising predictive performance on VWM (r = .402, p < .001), and identified features within the subcortical-cerebellum network, default mode network, motor network, corpus callosum, anterior corona radiata, and external capsule as significant predictors. The main results were then compared with those obtained on emotional regulation and fluid intelligence using the same pipeline, confirming the specificity of our findings. Moreover, the main results maintained well under different cross-validation regimes and preprocess strategies. These findings, while providing richer evidence for the importance of multimodality in understanding cognitive functions, offer a solid and general foundation for comprehensively understanding the VWM process from the top down.