Brain age has limited utility as a biomarker for capturing fluid cognition in older individuals.

One well-known biomarker candidate that supposedly helps capture fluid cognition is Brain Age, or a predicted value based on machine-learning models built to predict chronological age from brain MRI. To formally evaluate the utility of Brain Age for capturing fluid cognition, we built 26 age-prediction models for Brain Age based on different combinations of MRI modalities, using the Human Connectome Project in Aging (n=504, 36-100 years old). First, based on commonality analyses, we found a large overlap between Brain Age and chronological age: Brain Age could uniquely add only around 1.6% in explaining variation in fluid cognition over and above chronological age. Second, the age-prediction models that performed better at predicting chronological age did NOT necessarily create better Brain Age for capturing fluid cognition over and above chronological age. Instead, better-performing age-prediction models created Brain Age that overlapped larger with chronological age, up to around 29% out of 32%, in explaining fluid cognition. Third, Brain Age missed around 11% of the total variation in fluid cognition that could have been explained by the brain variation. That is, directly predicting fluid cognition from brain MRI data (instead of relying on Brain Age and chronological age) could lead to around a 1/3-time improvement of the total variation explained. Accordingly, we demonstrated the limited utility of Brain Age as a biomarker for fluid cognition and made some suggestions to ensure the utility of Brain Age in explaining fluid cognition and other phenotypes of interest.

Improving Predictability, Test-Retest Reliability and Generalisability of Brain-Wide Associations for Cognitive Abilities via Multimodal Stacking.

Brain-wide association studies (BWASs) have attempted to relate cognitive abilities with brain phenotypes, but have been challenged by issues such as predictability, test-retest reliability, and cross-cohort generalisability. To tackle these challenges, we proposed "stacking" that combines brain magnetic resonance imaging of different modalities, from task-fMRI contrasts and functional connectivity during tasks and rest to structural measures, into one prediction model. We benchmarked the benefits of stacking, using the Human Connectome Projects: Young Adults and Aging and the Dunedin Multidisciplinary Health and Development Study. For predictability, stacked models led to out-of-sample r ∼.5-.6 when predicting cognitive abilities at the time of scanning and 36 years earlier. For test-retest reliability, stacked models reached an excellent level of reliability (ICC>.75), even when we stacked only task-fMRI contrasts together. For generalisability, a stacked model with non-task MRI built from one dataset significantly predicted cognitive abilities in other datasets. Altogether, stacking is a viable approach to undertake the three challenges of BWAS for cognitive abilities.

Capturing brain-cognition relationship: Integrating task-based fMRI across tasks markedly boosts prediction and test-retest reliability.

Capturing individual differences in cognition is central to human neuroscience. Yet our ability to estimate cognitive abilities via brain MRI is still poor in both prediction and reliability. Our study tested if this inability can be improved by integrating MRI signals across the whole brain and across modalities, including task-based functional MRI (tfMRI) of different tasks along with other non-task MRI modalities, such as structural MRI, resting-state functional connectivity. Using the Human Connectome Project (n = 873, 473 females, after quality control), we directly compared predictive models comprising different sets of MRI modalities (e.g., seven tasks vs. non-task modalities). We applied two approaches to integrate multimodal MRI, stacked vs. flat models, and implemented 16 combinations of machine-learning algorithms. The stacked model integrating all modalities via stacking Elastic Net provided the best prediction (r = 0.57), relatively to other models tested, as well as excellent test-retest reliability (ICC=∼.85) in capturing general cognitive abilities. Importantly, compared to the stacked model integrating across non-task modalities (r = 0.27), the stacked model integrating tfMRI across tasks led to significantly higher prediction (r = 0.56) while still providing excellent test-retest reliability (ICC=∼.83). The stacked model integrating tfMRI across tasks was driven by frontal and parietal areas and by tasks that are cognition-related (working-memory, relational processing, and language). This result is consistent with the parieto-frontal integration theory of intelligence. Accordingly, our results contradict the recently popular notion that tfMRI is not reliable enough to capture individual differences in cognition. Instead, our study suggests that tfMRI, when used appropriately (i.e., by drawing information across the whole brain and across tasks and by integrating with other modalities), provides predictive and reliable sources of information for individual differences in cognitive abilities, more so than non-task modalities.

Variance and Scale-Free Properties of Resting-State Blood Oxygenation Level-Dependent Signal After Fear Memory Acquisition and Extinction.

Recently, the dynamic properties of brain activity rather than its stationary values have attracted more interest in clinical applications. It has been shown that brain signals exhibit scale-free dynamics or long-range temporal correlations (LRTC) that differ between rest and cognitive tasks in healthy controls and clinical groups. Little is known about how fear-inducing tasks may influence dispersion and the LRTC of subsequent resting-state brain activity. In this study, we aimed to explore the changes in the variance and scale-free properties of the brain's blood oxygenation level-dependent (BOLD) signal during the resting-state sessions before and after fear learning and fear memory extinction. During a 1-h break between magnetic resonance imaging (MRI) scanning, 23 healthy, right-handed volunteers experienced a fear extinction procedure, followed by Pavlovian fear conditioning that included partial reinforcement using mild electrical stimulation. We extracted the average time course of the BOLD signal from 245 regions of interest (ROIs) taken from the resting-state functional atlas. The variance of the BOLD signal and the Hurst exponent (H), which reflects the scale-free dynamic, were compared in the resting states before and after fear learning and fear memory extinction. After fear extinction, six ROIs showed a difference in H at the uncorrected level of significance, including areas associated with fear processing. H decreased during fear extinction but then became higher than before fear learning, specifically in areas related to the fear extinction network (FEN). However, activity in the other ROIs restored the H to its initial level. The variance of the BOLD signal in six ROIs demonstrated a significant increase from initial rest to the post-task rest. A limited number of ROIs showed changes in both H and variance. Our results imply that the variability and scale-free properties of the BOLD signal might serve as additional indicators of changes in spontaneous brain activity related to recent experience.

Brain Oscillatory Activity during Tactile Stimulation Correlates with Cortical Thickness of Intact Areas and Predicts Outcome in Post-Traumatic Comatose Patients.

In this study, we have reported a correlation between structural brain changes and electroencephalography (EEG) in response to tactile stimulation in ten comatose patients after severe traumatic brain injury (TBI). Structural morphometry showed a decrease in whole-brain cortical thickness, cortical gray matter volume, and subcortical structures in ten comatose patients compared to fifteen healthy controls. The observed decrease in gray matter volume indicated brain atrophy in coma patients induced by TBI. In resting-state EEG, the power of slow-wave activity was significantly higher (2-6 Hz), and the power of alpha and beta rhythms was lower in coma patients than in controls. During tactile stimulation, coma patients' theta rhythm power significantly decreased compared to that in the resting state. This decrease was not observed in the control group and correlated positively with better coma outcome and the volume of whole-brain gray matter, the right putamen, and the insula. It correlated negatively with the volume of damaged brain tissue. During tactile stimulation, an increase in beta rhythm power correlated with the thickness of patients' somatosensory cortex. Our results showed that slow-wave desynchronization, as a nonspecific response to tactile stimulation, may serve as a sensitive index of coma outcome and morphometric changes after brain injury.

Longitudinal changes of resting-state functional connectivity of amygdala following fear learning and extinction.

Altered functional connectivity of the amygdala has been observed in a resting state immediately after fear learning, even one day after aversive exposure. The persistence of increased resting-state functional connectivity (rsFC) of the amygdala has been a critical finding in patients with stress and anxiety disorders. However, longitudinal changes in amygdala rsFC have rarely been explored in healthy participants. To address this issue, we studied the rsFC of the amygdala in two groups of healthy volunteers. The control group participated in three fMRI scanning sessions of their resting state at the first visit, one day, and one week later. The experimental group participated in three fMRI sessions on the first day: a resting state before fear conditioning, a fear extinction session, and a resting state immediately after fear extinction. Furthermore, this group experienced scanning after one day and week. The fear-conditioning paradigm consisted of visual stimuli with a distinct rate of partial reinforcement by electric shock. During the extinction, we presented the same stimuli in another sequence without aversive pairing. In the control group, rsFC maps were statistically similar between sessions for the left and right amygdala. However, in the experimental group, the increased rsFC mainly of the left amygdala was observed after extinction, one day, and one week. The between-group comparison also demonstrated an increase in the left amygdala rsFC in the experimental group. Our results indicate that functional connections of the left amygdala influenced by fear learning may persist for several hours and days in the human brain.

Asymmetry of amygdala resting-state functional connectivity in healthy human brain.

Lateral asymmetry is one of the fundamental properties of the functional anatomy of the human brain. Amygdala (AMYG) asymmetry was also reported in clinical studies of resting-state functional connectivity (rsFC) but rarely in healthy groups. To explore this issue, we investigated the reproducibility of the data on rsFC of the left and right AMYG using functional MRI twice a week in 20 healthy volunteers with mild-to-moderate anxiety. We found a resting-state network of the AMYG, which included regions involved in emotional processing and several other brain areas associated with memory and motor inhibition. The AMYG network was stable in time and within subjects, but the right AMYG had more significant connections with anatomical brain regions. The rsFC values of the right AMYG were also more sustained across the week than the left AMYG rsFC. Subjective ratings of anxiety did not correlate significantly with the patterns of seed-based AMYG connectivity. Our findings indicate that, for healthy subjects, rsFC may differ for the right and left AMYG. Moreover, the AMYG functional connectivity is variable in short-term observations, which may also influence the results of longitude studies.

Correlation of BOLD Signal with Linear and Nonlinear Patterns of EEG in Resting State EEG-Informed fMRI.

Concurrent EEG and fMRI acquisitions in resting state showed a correlation between EEG power in various bands and spontaneous BOLD fluctuations. However, there is a lack of data on how changes in the complexity of brain dynamics derived from EEG reflect variations in the BOLD signal. The purpose of our study was to correlate both spectral patterns, as linear features of EEG rhythms, and nonlinear EEG dynamic complexity with neuronal activity obtained by fMRI. We examined the relationships between EEG patterns and brain activation obtained by simultaneous EEG-fMRI during the resting state condition in 25 healthy right-handed adult volunteers. Using EEG-derived regressors, we demonstrated a substantial correlation of BOLD signal changes with linear and nonlinear features of EEG. We found the most significant positive correlation of fMRI signal with delta spectral power. Beta and alpha spectral features had no reliable effect on BOLD fluctuation. However, dynamic changes of alpha peak frequency exhibited a significant association with BOLD signal increase in right-hemisphere areas. Additionally, EEG dynamic complexity as measured by the HFD of the 2-20 Hz EEG frequency range significantly correlated with the activation of cortical and subcortical limbic system areas. Our results indicate that both spectral features of EEG frequency bands and nonlinear dynamic properties of spontaneous EEG are strongly associated with fluctuations of the BOLD signal during the resting state condition.