2026-10-19

Dynamic functional network connectivity (dFNC) commonly identifies recurring connectivity states from resting-state fMRI, but states derived independently in each dataset lack guaranteed correspondence across studies and cannot be directly estimated in new participants. We introduce NeuroMark dFNC, which extends the NeuroMark principle of prior-guided network estimation to dynamic connectivity by using reproducible reference states as priors for estimating corresponding subject-specific states. Reference-state templates were derived from UK Biobank using windowed functional network connectivity and time-frequency moments (TFMs), with split-sample reproducibility evaluated for k-means, independent component analysis (ICA), and principal component analysis (PCA) template-generation approaches. Constrained ICA then used the templates to estimate subject-specific state patterns and continuous loadings in independent FBIRN schizophrenia and ADNI Alzheimer's disease cohorts. ICA loadings were calibrated by their contribution to reconstruction of the original feature matrix. Fractional occupancy and conditional mean dwell time were calculated from the state with the largest absolute calibrated loading at each observation. The predefined templates avoided the need for dataset-specific state discovery and post-hoc state matching. In both clinical cohorts, the analysis identified state-dependent, distributed edgewise differences and differences in selected temporal summaries. Nominal frequency-resolved findings varied across states and moment-frequency bins and were treated as exploratory. NeuroMark dFNC provides a common state space for cross-cohort comparison while retaining subject-specific connectivity patterns and continuous state expression.

Real time, chemically selective detection of neurotransmitter release in brain organoids remains a challenge due to their complex three dimensional architecture and limited optical accessibility. We introduce 3D printed carbon cup electrodes (CCEs) -pyrolyzed, multi level pillar structures - designed to confine a single organoid, increase electroactive surface area, and enable amperometric recordings of oxidizable neurotransmitters. The CCEs exhibit excellent electrochemical performance and reproducibility, supporting diffusion controlled dopamine redox behavior without interference from culture medium. Using ventral midbrain (vMB) organoids at two maturation stages (Day 30 and Day 90), we demonstrate real time detection of potassium evoked dopamine release directly from organoids. Mature organoids exhibit higher peak currents, correlating with increased tyrosine hydroxylase (TH) expression confirmed by immunocytochemistry. Embedding organoids in Matrigel significantly attenuates dopamine signals, indicating a diffusion barrier that limits stimulant access and/or dopamine transport to the electrode. Repetitive stimulations reveal progressive depletion of dopamine stores, consistent with diminished vesicular release. Chemical perturbations using 6 hydroxydopamine and verapamil further validate the platform by producing substantial reductions in dopamine release accompanied by loss of TH neurons. These results establish CCEs as a powerful tool for label free, real-time monitoring of neurotransmission in human organoids, with potential for multiplexed sensing and integration of enzymatic detection strategies for non electroactive neurotransmitters.

Spectral modeling of electrophysiological recordings enables non-invasive characterization of neural excitability. However, current approaches rely on a single spectral index, limiting insight into the multiple biophysical processes underlying excitation-inhibition balance. Here, we introduce a biologically informed, multi-parametric spectral analysis that disentangles neural excitability and synaptic kinetics from recordings. As a proof of concept, we demonstrate that this approach captures signatures of hyperexcitability and glutamate reuptake dysfunction characteristic of Alzheimer's disease.

The role of Rapid Eye Movement (REM) sleep in memory consolidation remains poorly understood. Several influential ideas propose that REM sleep serves to eliminate crosstalk between memories, thereby orthogonalizing representations of similar ones; however, mechanistic confirmation remains sparse. Here we used recurrent neural networks trained with Equilibrium Propagation (EP), a biologically plausible training algorithm, to investigate the role of REM-like sleep in memory consolidation. We found that REM-like sleep reduces memory interference by decorrelating task-specific neuronal populations, isolating individual memory traces. During sleep, reactivation was dominated by the population encoding the most recent memory. Net excitatory input onto this population decreased, most strongly that from older-memory populations, which allowed inputs of older memories to drive their own neurons again. Spreading the same total synaptic change evenly across all target populations abolished this recovery. We call this mechanism reactivation-targeted pruning: plasticity is confined to one population, whose neurons lose inputs that are silent when they fire. The mechanism held under Dale's law, and depression of excitatory synapses alone, without any inhibitory plasticity, was sufficient. These findings suggest mechanistic principles underlying REM sleep's role in memory consolidation and highlight the potential for integrating human-like learning behaviors into artificial neural networks.