NeuroImage

Multi-site diffusion MRI studies face scanner-induced variability that can obscure biological signal. Harmonization methods such as ComBat have been developed to address this, but have been evaluated primarily on diffusion scalar metrics. Whether scanner reproducibility differs across fundamentally distinct tract-derived representations has not been systematically compared. Here, we compared the reproducibility of three metric families (diffusion, shape, and connectivity) across 36 association tracts using the MASiVar dataset (5 subjects, 4 scanners, 27 sessions). We assessed intraclass correlation coefficients (ICC) and multivariate subject discrimination at baseline, under dimensionality reduction, and after ComBat harmonization. At baseline, shape metrics showed the highest reproducibility (median ICC 0.69), followed by connectivity (0.49) and diffusion (0.34). Shape and connectivity achieved comparable subject discrimination (both 1.75), significantly exceeding diffusion (1.23). ComBat harmonization improved all families but harmonized diffusion (0.58) remained below unharmonized shape (0.69), indicating that metric family selection remains consequential even after harmonization. Under low-dimensional representation, connectivity showed the largest gains (ICC 0.86, subject discrimination 3.0), exceeding other families at any dimensionality. Analysis of principal component loadings identified a small number of cortical regions per tract (median 6) that capture 95% of the reproducible connectivity signal, providing a per-tract reference for selecting the most informative regions in future multi-site studies. These findings indicate that the choice of which tract-derived metrics to analyze in multi-site studies deserves at least as much consideration as how to harmonize them.

Harmonisation is widely used to mitigate site- and scanner-related batch variability in multisite neuroimaging studies and is particularly critical in longitudinal clinical trials, where detection of subtle biological or treatment-related changes depends on reliable measurement across scanners and timepoints. However, the effectiveness of harmonisation in small, heterogeneous clinical datasets remains insufficiently understood, particularly in relation to subject-level variability and consistency across acquisition settings, and its impact on both removal of technical variability and preservation of biological variation in pooled multisite analyses. We systematically evaluated a range of image-based and statistical harmonisation methods using a clinically realistic multisite, multiscanner structural T1-weighted (T1w) MRI test-retest dataset comprising three controlled acquisition scenarios: repeatability, intra-scanner reproducibility and inter-scanner reproducibility. Methods were applied under different batch specifications (site, scanner, or both) and performance was assessed within each scenario and in pooled data using a multi-metric framework capturing both technical and biological variability in volumetric imaging-derived phenotypes (IDPs) relevant to aging and dementia research. Across IDPs, before harmonisation variability was lowest in the repeatability scenario (median variability=0.6 to 2.7%, rank consistency {rho} [≥]0.9), with modest increases under intra-scanner reproducibility (0.5 to 3.2%, {rho}=0.5 to 1.0) and substantially greater variability under inter-scanner reproducibility conditions (1.7 to 19.2%, {rho} =-0.1 to 0.9). These results offer important information to consider for multisite study design, including sample size calculation in clinical trials. Harmonisation performance was strongly context dependent, with clearer benefits emerged in inter-scanner scenarios where both variability reduction and improvements in subject-level consistency were observed. In pooled data, approaches that explicitly modelled site as batch and accounted for repeated-measure structure showed greater consistency across IDPs in batch effect mitigation and more accurately reflected underlying biological variation. Our evaluation metrics enabled disentangling the removal of global batch effect while highlighting residual variability at the phenotype-specific or multivariate levels. These findings demonstrate that harmonisation cannot be treated as a one-size-fits-all solution and must be interpreted relative to the acquisition context, dataset structure, and downstream analytic goals. Multi-metric evaluation under realistic clinical constraints is essential to support reliable and translatable neuroimaging inference by ensuring appropriate correction of batch effects while preserving longitudinal biological signals and sensitivity to clinically meaningful change in multisite studies.

The Adolescent Brain Cognitive Development (ABCD) Study is the largest U.S.-based neuroimaging initiative of adolescent brain maturation. Diffusion MRI (dMRI) provides unique insights into white matter organization, yet applying advanced processing pipelines and managing technical variability across scanning environments remains challenging at scale. To address these issues, we present ABCD-BIDS Community Collection (ABCC) release 3.1.0, including a curated resource of more than 24,000 fully processed ABCD dMRI datasets. ABCC provides fully processed images, nuanced image quality metrics, advanced microstructural measures, and person-specific bundle tractography. Evaluating these rich data revealed that measures of diffusion restriction and non-Gaussianity--in particular the intracellular volume fraction from NODDI and return-to-origin probability from MAP-MRI--were highly sensitive to neurodevelopment and robust to variation in image quality. Additionally, harmonization of microstructural features markedly improved the cross-vendor generalizability of developmental effects. Together, ABCC accelerates reproducible, rigorous research on adolescent white matter development.

Recent advances have shown that the complexity of neural signals tracks global states of consciousness, such as wakefulness versus sleep. However, it is still unclear to what extent neural complexity reflects fine-grained changes in conscious content within the same global state. Here, we investigate how the complexity of brain signals is affected by increased perceptual clarity of a stimulus. To this end, we estimated neural signal complexity using Complexity via State-space Entropy Rate (CSER) to EEG recordings from an auditory discrimination task. In this paradigm, auditory stimuli were presented at varying signal-to-noise ratios (SNRs), with higher SNRs corresponding to greater subjective audibility and perceptual clarity, enabling us to relate neural complexity to graded perceptual awareness within a constant global state of consciousness. Our results showed that, while broadband CSER remains constant across SNRs, its spectral decomposition displays frequency-specific effects, with higher SNRs associated with a decreased complexity in and {beta} bands, increased complexity in{delta} , and no significant changes in{gamma} . Additionally, a temporal investigation of CSER exhibited a significant increase in complexity with stimulus clarity, with deviations from baseline peaking approximately 30 ms before the ERP. Extending this analysis to pairs of brain regions, mutual information rate uncovered a sudden post-stimulus breakdown in long-range information transmission relative to baseline. Taken together, these results reveal that while aggregated complexity measures track global states of consciousness, time- and frequency-resolved information-theoretic measures can capture variations in perceptual awareness, demonstrating their sensitivity as estimators of the level of conscious experience.

Salient sensory stimuli are known to evoke neural activations across distributed brain regions. However, the temporal dynamics of these responses over sub-second timescales remain poorly understood, in part due to limitations in the temporal resolution of non-invasive neuroimaging methods. We examined the spatiotemporal dynamics of neural activations evoked by salient sensory stimuli (rare sounds) using 1,194 widely distributed intracranial electrodes in 5 neurosurgical patients. Salient stimuli preferentially activated 263 of 1,194 electrodes (22%), with responses segregating into two largely distinct spatiotemporal patterns: (1) phasic activation in sensorimotor regions, and (2) sustained activation within the salience network. Cross-correlation analysis revealed that phasic sensorimotor activation preceded sustained salience network activation on a trial-by-trial basis. These findings support an updated view of salience processing in the human brain, revealing that salient stimuli evoke two sequential stages of neural activation--phasic sensorimotor responses followed by sustained salience network activity--rather than simultaneous widespread activation.

Brain network analysis has emerged as a critical framework for understanding the complex organization and function of the human brain, underpinning insights into cognition, behavior, and neuropsychiatric conditions. Central to this approach is the parcellation of the brain into discrete regions, which simplifies high-dimensional connectome data and facilitates the investigation of network architectures. However, the proliferation of brain parcellation schemes introduces significant challenges: different parcellations often yield varying network sizes and measures, complicating cross-study comparisons and the reproducibility of findings. Moreover, most connectome construction pipelines are rigid, typically outputting connectivity matrices from only one or a few parcellation schemes, which limits flexibility. In this paper, we address these issues by introducing BridgeBP, a novel toolbox designed to bridge brain parcellations by leveraging continuous brain connectivity concepts. BridgeBP transforms structural connectivity matrices derived from one parcellation scheme into matrices corresponding to more than 40 alternative schemes, standardizing analyses and enhancing the robustness of network studies. Through extensive evaluations, we demonstrate that BridgeBP enables consistent network comparisons across diverse parcellation frameworks, paving the way for more reproducible and generalizable insights in brain connectome research.

1.Quantitative neuropathology has advanced through whole-slide imaging and digital histology platforms. Yet, these measurements rarely align with neuroimaging coordinate frameworks that may be useful for spatial modeling and other applications. QNPtoVox, short for quantitative neuropathology to voxels, is a reproducible, modular pipeline that transforms quantitative metrics generated by digital pathology software (HALO) into voxel-based maps registered to a standard common coordinate (MNI) template. The workflow integrates digital histopathology, gross tissue photography, ex-vivo MRI, and nonlinear registration to generate spatially standardized 3D pathology representations. This Methods article provides a complete procedural description, including required materials, step-wise instructions, operator-dependent checkpoints, expected outputs, reproducibility evaluation, and troubleshooting. QNPtoVox enables voxel-level integration of neuropathology with neuroimaging tools, unlocking existing histopathology datasets for computational modeling and cross-cohort harmonization.

Hidden (semi-) Markov Models (HsMMs) are increasingly being used to segment neurophysiological signals into sequences of latent cognitive processes. The idea: different processes will leave distinct traces in trial-level recordings of (multivariate) neuro-physiological signals. Markov models, equipped with an emission model of these traces and a latent process model describing the progression through the different latent processes involved in a task, can then be used to infer the most likely process for any time-point and trial. However, the currently used HsMMs remain limited in two important ways. First, they cannot account for subject-level heterogeneity in the latent and emission process. Instead, a single group-level model is assumed to explain the entire data. Second, they cannot account for the potentially non-linear effects of experimental covariates on the latent and emission process. To address these problems, we present a modeling framework in which the HsMM parameters of the emission and latent process are replaced with mixed additive models, including smooth functions of experimental covariates and random effects. We derive all necessary quantities for empirical Bayes and fully Bayesian inference for all parameters and provide a Python implementation of all estimation algorithms. To demonstrate the advantages offered by this framework, we apply such a multi-level model to an existing lexical decision dataset. We show that, even in such a simple task, not all subjects rely on the same processes equally and that at least two semi-Markov states, previously believed to reflect distinct processes, might actually relate to the same cognitive process.

Short-interval intracortical inhibition (SICI) is the most widely used neurophysiological index of GABAergic inhibition in the human cortex. However, it is an indirect measure, inferring synaptic inhibition from suppression of peripherally recorded motor-evoked potentials (MEPs) elicited by transcranial magnetic stimulation (TMS). In the standard protocol, a subthreshold conditioning pulse suppresses the MEP evoked by a suprathreshold test pulse delivered 1-5 ms later. Interpretation is further complicated by temporal overlap with short-interval intracortical facilitation (SICF), reflecting excitatory interactions at interstimulus intervals of [~]1.5 and 2.7 ms. To overcome these limitations, we recorded immediate TMS-evoked EEG potentials (iTEPs; 1-10 ms post-stimulus) as a more direct measure of motor cortical activity in 16 healthy volunteers (20-35 years; 7 male). The conventional SICI protocol suppressed only later components of the iTEP, likely corresponding to late corticospinal volleys previously identified in epidural spinal recordings after suprathreshold TMS, while the earliest iTEP component was unaffected. Importantly, later iTEPs were suppressed to a similar extent whether conditioning-test intervals coincided with SICF peaks or troughs, and the magnitude of iTEP suppression correlated with concurrently recorded paired-pulse MEP suppression. SICI also reduced an early TEP component (N15; 10-20 ms), but paired-pulse N15 suppression showed a different dependence on stimulus intensity and did not correlate with MEP suppression. These findings demonstrate that SICI measured via MEPs does not reflect a global index of cortical GABAergic motor cortical inhibition but instead reflects inhibition within specific cortical circuits that can be investigated directly with iTEPs.

BackgroundPsilocybin is a classic psychedelic that acutely alters brain functional connectivity. These changes are linked to therapeutic doses and subjective effects, with some evidence that changes persist beyond acute drug administration. However, the effects of lower doses on sustained connectivity changes remain unclear. MethodsTen healthy volunteers received three psilocybin doses (between 5 and 20 mg) in a randomized and blinded order, with at least one week between doses. Resting-state functional magnetic resonance imaging was completed at baseline and one week after a single dose. Functional connectivity changes were analyzed in relation to dose and altered conscious states at both the level of individual brain region connections (edges) and resting-state networks. ResultsDose-dependent changes in 77 edges (76 increases, 1 decrease, of 1275 possible) were observed, but none survived multiple-comparison correction. At the network level, we observed one dose-dependent between-network increase (of 21 possible), and one dose-dependent within-network increase (of seven possible); the latter surviving correction. Alterations in conscious state were positively associated with widespread connectivity changes (dose-adjusted), with many network-level associations surviving correction. These directional patterns showed that lower doses and smaller conscious state alterations were linked to decreased connectivity, whereas higher doses and greater conscious state alterations were linked to increased connectivity. ConclusionsDose level and acute subjective effects were positively associated with multiple functional connectivity changes one week after a low-to-moderate psilocybin dose. Further research is warranted to characterize these sustained effects and their therapeutic relevance to inform studies adopting similar dosing regimens in clinical cohorts. Trial RegistrationAustralian New Zealand Clinical Trials Registry: ACTRN12621000560897 Date registered: 12 May 2021 URL: https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=381526&isReview=true

Multimodal machine learning leverages complementary information from diverse data sources and has shown strong promise in medical imaging, where multimodal data is critical for clinical decision making. In glioma grading, integrating MRI modalities with clinical data can improve diagnostic accuracy, yet systematic comparisons of fusion strategies remain limited. This study evaluates early, intermediate, and late fusion approaches, addressing the question: How does the inclusion of clinical data alongside MRI modalities influence grading performance? To assess modality contributions, we design adaptable fusion layers and employ interpretability techniques, including attention-based analysis. Our results show that incorporating clinical data consistently outperforms unimodal and MRI-only baselines, with intermediate fusion yielding the most reliable gains. Beyond accuracy, the framework reveals how MRI and clinical features jointly shape predictions, underscoring the importance of both fusion design and interpretability for clinical adoption.

Understanding neural correlates of brain function in neuroscience now largely involves detecting and analyzing transient signals from fluorescent sensors. Imaging technologies such as confocal and two-photon microscopy, along with onboard miniscopes, enable the visualization of neural activities and capture dynamic signals both ex vivo and in vivo. This includes monitoring Ca2+ transients via the expression of genetically encoded sensors such as GCaMP in specific brain cells. Additionally, the advent of GPCR-based neurotransmitter sensors allows for imaging the release of neurotransmitters including glutamate and GABA, as well as neuromodulators such as dopamine or noradrenaline. These approaches however generate large, high-dimensional, spatiotemporally complex datasets, presenting significant challenges for signal detection and analysis. To overcome these challenges, we developed a versatile pipeline of Dynamic Extraction and Tracking of Emitted Cellular Transients (DETECT), which combines background denoising, object segmentation, and multi-object tracking. Our user-friendly, Python-based GUI offers a low-resource platform for efficient data analysis. Validated across various imaging modalities and biological models, DETECT provides a robust and comprehensive solution for analyzing complex imaging datasets in neuroscience research.

Early behavioral and temperamental differences are important indicators of later socioemotional development and psychopathology risk, yet their neural bases near birth remain incompletely understood. Using resting-state fMRI data from the Developing Human Connectome Project, we examined whether neonatal functional connectivity predicts 18-month behavioral and temperament outcomes in 397 infants (277 term-born, 120 preterm-born). Outcomes were assessed using the Child Behavior Checklist (CBCL) and the Early Childhood Behavior Questionnaire (ECBQ). We applied a stability-driven, ROI-constrained connectome-based predictive modeling framework to identify robust whole-brain connectivity features associated with later externalizing, internalizing, surgency, negative affect, and effortful control. Significant predictive models were observed for multiple outcomes across the whole cohort as well as within term-born and preterm-born groups, with clear differences in predictive architecture between cohorts. Across analyses, prefrontal and temporoparietal systems were repeatedly implicated, alongside medial temporal, fusiform, parahippocampal, and orbitofrontal-related regions. These findings indicate that large-scale neonatal functional organization is meaningfully related to later socioemotional and behavioral variation, and that preterm birth is associated with partly distinct predictive connectivity patterns.

ObjectiveSocial difficulties are transdiagnostic in childhood, but their heterogeneity is poorly characterised and rarely treated as a primary neurodevelopmental phenotype. This matters because childhood and adolescence are sensitive periods for peer relationships and brain development. We used data-driven modelling and non-linear mapping to derive social profiles and test their clinical, cognitive, and neural correlates. MethodsParticipants were 992 children aged 5-18 years from CALM (Mage = 9.6). Social items from the SDQ, CCC-2, and Conners-3 were modelled using a regularised partial correlation network to derive core social dimensions. A self-organising map captured graded social profiles. Simulated archetypes, SVM-based island identification, and permutation testing defined profile regions and centroid-distance scores. Profiles were related to referral, diagnosis, cognition, BRIEF indices, and T1-derived MIND network structure in an MRI subsample (n = 431). ResultsWe identified four profiles: social engagement, friendship difficulties, social withdrawal, and peer victimisation. Profile expression tracked variation in referral and diagnostic pathways. Social withdrawal showed the clearest disadvantage across cognitive domains, whereas social engagement was associated with fewer executive function difficulties across BRIEF indices. MIND strength components covaried with profile expression (a significant PLS latent variable, p = 0.02), with covariance strongest for social withdrawal and peer victimisation. ConclusionsChildhood social functioning organises graded signatures that relate to clinically relevant pathways, cognitive and executive outcomes, and brain structure. Profiling social signatures provides a scalable framework for identifying social need beyond diagnostic categories, motivating studies to test directionality and improve developmental outcomes.

Electroencephalography (EEG) preprocessing is a critical yet time-consuming step that often relies on expert-driven, semi-automatic pipelines, limiting scalability and reproducibility across large datasets. In this work, we present sEEGnal, a fully automated and modular pipeline for EEG preprocessing designed to produce outputs comparable to expert-driven analyses while ensuring consistency and computational efficiency. The pipeline integrates three main modules: data standardization following the EEG extension of the Brain Imaging Data Structure (BIDS), bad channel detection, and artifact identification, combining physiologically grounded criteria with independent component analysis and ICLabel-based classification. Performance was evaluated against manual preprocessing performed by EEG experts at two complementary levels: preprocessing metadata (bad channels, artifact duration, and rejected components) and EEG-derived measures. In addition, test-retest analyses were conducted to assess the stability of the pipeline across repeated recordings. Results show that sEEGnal achieves performance comparable to expert-driven preprocessing while preserving key neurophysiological features. Furthermore, the pipeline demonstrates reduced variability and increased consistency compared to human experts. These findings support sEEGnal as a robust and scalable solution for automated EEG preprocessing in both research and large-scale applications. HighlightsFully automated and modular EEG preprocessing pipeline. Benchmarked against expert-driven preprocessing. Comparable performance in metadata and EEG-derived measures. Demonstrates stable performance in test-retest recordings. BIDS-based framework for reproducible EEG data handling.

Focal cortical dysplasias (FCDs) are one of the most common structural causes of drug-resistant epilepsy in children but are frequently subtle and difficult to detect on conventional MRI. Many automated lesion detection methods have therefore been proposed to support neuroradiological assessment. In this study, we externally validated two recently developed deep-learning approaches for FCD detection, MELD Graph and 3D-nnUNet, in a pediatric cohort. In this retrospective single-center study, brain MRI scans of 71 children evaluated for epilepsy were analyzed, including 35 MRI-positive patients with suspected FCD and 36 MRI-negative cases based on the primary radiology reports. Both models were applied to standard 3D T1-weighted and 3D FLAIR images. Detected lesions were reviewed by an experienced pediatric neuroradiologist and classified as true positive, false positive, or false negative. Clinical semiology and EEG findings were additionally evaluated for cases with false-positive detections. At the lesion level, MELD Graph achieved a precision of 0.85 and recall of 0.52, while 3D-nnUNet achieved a precision of 0.91 and recall of 0.48. In the MRI-negative patients, MELD Graph produced more false-positive detections than 3D-nnUNet (0.53 vs. 0.14 false-positive lesions per patient). At the patient level, MELD Graph showed slightly higher sensitivity than 3D-nnUNet (0.63 vs. 0.54), whereas 3D-nnUNet demonstrated markedly higher specificity (0.86 vs. 0.56). Improved FLAIR image quality was associated with trends toward improved model performance. Both models demonstrated high precision but moderate sensitivity, indicating that they are valuable decision-support tools but cannot replace expert neuroradiological evaluation. Optimized MRI acquisition protocols are needed to further improve automated lesion detection in pediatric epilepsy.

Cognitive reappraisal, the deliberate reinterpretation of emotional events, is widely considered an effective emotion regulation strategy, and modulation of the late positive potential (LPP) during negative affect reduction has become the primary electrophysiological evidence for volitional emotional control. Experimental instructions, however, impose dual-task demands that free viewing does not, confounding reappraisal with cognitive load. By including instructions to increase emotional responses to pictures ("enhance") as well as instructions to decrease ("suppress"), different predictions are generated. If the LPP reflects regulation, then, compared to free viewing, suppress instructions should decrease LPP amplitude, and enhance instructions should increase LPP amplitude. If modulation instead reflects cognitive load, both instructions should reduce the LPP, as both impose an additional cognitive task. In a sample of 107 participants, evaluative ratings confirmed that regulation instructions modulated reported emotional intensity in the expected directions (Enhance > View > Suppress), but that both enhance and suppress instructions reduced LPP amplitude compared to free viewing, with Bayesian model comparisons providing strong evidence against direction-specific regulation and in favor of cognitive load. Whole-scalp multivariate pattern analysis confirmed that no instruction-related neural signal exists at any scalp location or latency within the first second after stimulus onset. These data indicate that LPP modulation following both instruction types reflects dual-task cognitive load rather than volitional emotional control. Significance StatementCognitive reappraisal is considered the gold standard of emotion regulation, and reduced late positive potential (LPP) amplitude during negative emotion suppression is the primary neural evidence that humans can voluntarily control emotional responses. The current data are inconsistent with this regulatory account and instead support a cognitive load interpretation. Whether instructed to enhance or suppress emotional responses, LPP amplitude was reduced in both conditions relative to free viewing, consistent with attentional resource competition rather than directional regulatory control. The same participants reported successfully regulating emotional experience in opposite directions, producing a clear dissociation between neural and behavioral measures. These findings challenge a basic tenet of emotional regulation and raise questions concerning LPP modulation as a biomarker of regulatory capacity.

Myelin is a highly complex membranous structure wrapped around axons by oligodendrocytes or Schwann cells in the central and peripheral nervous system, respectively. Fluorescent labeling is widely used to study the structure and dynamics of myelin. Combining structural with functional imaging requires labeling of myelin with red fluorescence, as many functional sensors, including Ca2+ indicators and genetically encoded metabolite sensors, fluoresce in the green spectral range. However, in vivo tools enabling red fluorescent labeling of myelinating cells and their myelin sheaths remain limited. Here, we generated a set of seven transgenic mouse lines expressing a membrane-targeted variant of the red fluorescent protein tdTomato in myelinating oligodendrocytes and Schwann cells throughout the nervous system. The mouse lines provide a variety of expression patterns ranging from wide-spread labeling of myelin to a rather sparse expression, the latter enabling visualization of individual oligodendrocytes and their associated myelin sheaths. In the peripheral nervous system, the pattern of fluorescence in sciatic nerves indicates predominant localization of tdTomato to non-compact myelin compartments including the inner and outer tongues, paranodal loops and Schmidt-Lanterman incisures. In summary, our work provides a set of novel mouse lines with myelin labeled by red fluorescence, which are compatible with diverse imaging modalities in the green spectral range enabling integrated structural and functional imaging. Main PointsO_LITransgenic mouse lines expressing membrane-targeted tdTomato in myelin enable imaging of myelin in the red spectral range C_LIO_LIDistinct expression patterns range from wide-spread labeling to sparse single-cell resolution, supporting diverse imaging applications C_LI

ObjectiveIn individuals with drug-resistant epilepsy, accurately identifying the brain regions where seizures originate is a critical prerequisite to guide surgical treatment and achieve seizure freedom. To accomplish this, intracranial EEG is considered the gold standard, providing the spatiotemporal high-resolution data necessary to pinpoint epileptogenic activity. However, this precision is achieved through an invasive procedure with significant patient burden, which is fundamentally limited by the electrode placement and spatial coverage. MethodsIn this study, we investigated the potential utility of preoperative resting-state fMRI to non-invasively map alterations in brain dynamics at the whole brain level. Region-wise brain dynamics were quantified with complementary measures of local autocorrelation decay rates. We assessed the capacity of these derived features to effectively identify intracranial EEG confirmed seizure onset zones in 18 individuals with drug-resistant medial temporal lobe epilepsy. Overall, the study cohort contained 3867 implanted electrodes of which 159 classified as seizure onset zones by two independent board-certified epileptologists. ResultsOverall, our findings reveal more constrained temporal dynamics for brain regions associated with seizure onsets compared to non-seizure onset zones. Individual-level prediction showed a performance better than chance in 15 of the 18 patients. The overall predictive performance across all patients yielded a median AUC of 0.81, a median true positive rate of 0.75, and a median true negative rate of 0.83. Furthermore, in a subset of 13 patients, those with negative seizure outcomes showed higher probabilities of seizure onset zone predictions outside the resection area compared to those with good outcomes. SignificanceOverall, our findings suggest that altered temporal dynamics derived from preoperative resting-state fMRI represent a promising non-invasive approach for delineating epileptogenic tissue, potentially informing intervention strategies and guiding electrode placement.

Human sentence comprehension proceeds word-by-word, with prior research proposing two central sources of cognitive demand during incremental processing: forward-looking disambiguation of the incoming information stream, and backward-looking retrieval of information associated with previous words from working memory. Recent work has shown that Transformer-based language models successfully generate predictions about sentence processing load in human psycho- and neurolinguistic data by operationalizing disambiguation cost as next-token surprisal, and memory retrieval cost as normalized attention entropy (NAE). Such models, however, remain difficult to interpret as it is not well understood what factors play causally into the decision to assign a cost value to a given word in such artificial neural networks. Here, we present interpretable and cognitively grounded models of disambiguation and memory retrieval and evaluate their neural alignment and spatio-temporal correlates using human magnetoencephalography responses to naturalistic narrative speech. Multivariate temporal response function modeling demonstrates firstly that these human-bias-informed models fare equally well in accounting for observed human language processing data as their Transformer counterparts. This same modeling framework then suggests that surprisal and NAE temporally dissociate in the cortical language network -- surprisal being predictive of bilateral superior temporal gyrus and supramarginal gyrus activation [~]300-500 ms, and NAE being predictive of activity in the same regions, but later [~]750-850 ms. By demonstrating that interpretable neurocomputational models can achieve meaningful brain alignment while maintaining explanatory transparency, this work offers a methodological blueprint for bridging the gap between algorithmic theory and neural implementation.