This study characterizes 2D spatial profiles of mesoscale functional columns in the primary somatosensory cortex using high-field fMRI and AI-based clustering. Findings reveal distinct spatial features between cortical layers and high correspondence between stimulus and resting states, offering new insights into layer-specific functional organization and data-driven neuroimaging applications.

Five machine-learning models were trained to classify autism using ABIDE. Similar performance suggested accuracy variations in the literature may stem from inclusion criteria, modalities, and evaluation pipelines. SmoothGrad investigated the stability of features identified by models and showed that structural and functional features from the ventricles and temporal cortex contribute to autism identification.

Our study reveals a significant relationship between functional connectivity and underlying microstructure during development that decreases with age and varies across connection types. Despite the progressive maturational decoupling of microstructural and functional connectivity, a shared network structure may underlie changes in both properties.

We aimed to determine the effect of alterations in average controllability (AC) on emotional and cognitive symptoms in patients with major depressive disorder (MDD). We found that AC values in the right SFG were negatively moderating the association between memory (i.e., DSF scores) and anxiety symptoms (i.e., HAMA scores) in patients with MDD.

HIMA, a novel multiple imputation method specifically designed for high-dimensional neuroimaging data, drastically reduces computational burden (e.g., 1 h vs. 800 h for traditional methods) while improving imputation precision and stability, as evidenced by theoretical justification, extensive simulations, and real data analysis.

Neural oscillations of the alpha and beta bands play a crucial role in the human brain's integration of internal and external reference frames.

White matter (WM) health declines with age and improves with estimated fitness. However, the effects of age and estimated fitness do not completely overlap spatially. Cerebrovascular health mediates the relationships of WM health with both age and estimated fitness. Males show steeper declines in WM health than females.

This study demonstrates that regular musical rhythms can act as a temporal guide for brain oscillations, enhancing synchronization with the speech envelope. This effect was observed across different rhythmic structures and languages, suggesting that the regularity in musical rhythms plays a crucial role in improving cortical tracking of speech.

Functional and structural connectivity data was used to predict verbal fluency and vocabulary knowledge in a lifespan sample using two machine learning algorithms. The predictability of verbal compared to nonverbal abilities in younger and older adults was found to be low across whole-brain and network-specific functional and structural imaging data.

We investigated if white matter shape could predict cognitive performance. Using explainable machine learning, we trained 210 models to predict cognitive performance. Most shape measures are as predictive as microstructure or connectivity information. Results suggest the shape of the brain's fiber tracts is important for future study.

We demonstrate that prediction violations modulate steady-state activity at 40 Hz. The degree of modulation was associated with the degree of myelin in the cerebellum as measured with the T1w/T2w ratio.

In this study, 7 brain atlases, 6 brain disorders, and 3 analysis methods were systematically compared and analyzed. Our cross-atlas/disorder analyses indicate that frontal-related FCs may serve as potential common and robust neuro-abnormalities across 6 psychiatric disorders. Furthermore, in order to improve the replicability of rsfMRI-based FC analyses, this study suggests the use of functional templates at larger granularity.

We explored whether and how effective motor skill learning modulates load-dependent patterns in continuous hand movements. Our findings provide robust evidence of a learning-induced inverted-U pattern in the left (contralateral) Pre-SMA using linear mixed-effect models. The emergence of the inverted-U pattern after training appears to be driven by decreased activity at the lowest and highest task loads.

MEG imaging of task-related activity during a “go”/“no-go” task revealed that pediatric post-stroke patients with dystonia had increased theta power during correct “No-go” trials, and reduced gamma power and lower gamma peak frequency during correct “Go” responses, compared to non-dystonic post-stroke patients and healthy children.

We propose a robust deep learning framework, TractCloud-FOV, to achieve robust and effective tractography parcellation in diffusion MRI with an incomplete field of view by introducing a synthetic data augmentation strategy to simulate real-world inferior FOV cutoff scenarios.

This study investigates the integration of static and dynamic structure–function coupling (SFC) for early Alzheimer's disease (AD) classification. By combining machine learning with SFC metrics, it reveals how AD alters brain network stability, offering new insights for early AD detection and pathophysiological understanding.

Results demonstrate converging evidence of age-differential patterns of IFG and whole brain rsFC associated with PI in working memory, with several regions previously implicated in the ability to control PI in working memory showing differential rsFC patterns in older and younger adults, both cross-sectionally and longitudinally.

Delivering stimulation to the correct target is a challenge for deep brain stimulation. We propose a new approach that allows condensing a large battery of scores onto a few main dimensions, identifying the structural connections associated with each dimension, and estimating changes in individual scores based on their stimulation.

We use the resting-state fMRI data from a large cohort to reveal tentative transcranial magnetic stimulation targets at cortical regions within the hippocampal network. We identified reproducible individualized targets at the medial prefrontal cortex and posterior parietal cortices. Our results suggest candidate TMS target coordinates to modulate the hippocampal function.

We analyzed and integrated data from two separate types of brain scans, fMRI and MEG, from 15 healthy human participants on the psychedelic drug LSD. Our work identified changes in the connectivity of a frontal brain region, which relate to two common experiences on LSD: emotional arousal and ego dissolution.

This study reveals that Subjective Cognitive Decline (SCD) is characterized by dynamic, state-dependent connectivity changes in the amygdalohippocampal complex. These changes highlight compensatory and synchronized brain states, with implications for understanding early cognitive decline and informing potential biomarkers and intervention strategies.

Lower quantitative susceptibility in white matter is associated with a greater number of prior concussions, worse symptoms, and worse neurocognitive function in healthy male and female collegiate-aged athletes. Results suggest that subtle differences in brain microstructure due to repeated concussions might partially underlie symptom complaints and cognitive performance in athletes.