2025 Hyper Recent •CC0 1.0 Universal

This work is dedicated to the public domain. No rights reserved.

Access Preprint From Server
April 20th, 2025
Version: 1
CEA/NeuroSpin, France
neuroscience
biorxiv

Toward Unified Biomarkers for Focal Epilepsy

Wang, S. H.Open in Google Scholar•Ferrari, P.Open in Google Scholar•Arnulfo, G.Open in Google Scholar•Marzulli, M.Open in Google Scholar•Degras, D.Open in Google Scholar•Myrov, V.Open in Google Scholar•Palva, S.Open in Google Scholar•Nobili, L.Open in Google Scholar•Ciuciu, P.Open in Google Scholar•Palva, M.Open in Google Scholar

Surgical resection of the epileptogenic zone (EZ) is the most effective treatment for drug-resistant focal epilepsy. However, 30[ndash]80% of patients do not attain long-term seizure freedom post-surgery, indicating that pre-surgical evaluation often falls short of fully delineating the EZ. Accurate EZ-localization is challenging because each individual EZ is a unique and complex network, often comprising multiple overlapping neuropathologies. This hypothesis is supported by studies showing that combining multiple biomarkers enhances localization accuracy. Nonetheless, combining biomarkers leads to high-dimensional features that risks overfitting classifiers and reduces the interpretability of underlying neuropathology, ultimately limiting clinical applicability. We asked whether high-dimensional neuronal features could be reduced to a low-dimensional latent space to captures essential characteristics of the epileptogenic network (EpiNet). In the latent space, each sampled brain region is assigned an EpiNet-saliency score along a 0[ndash]1 continuum, characterizing the degree of epileptogenic proneness from lowest to highest. Using 10-minute interictal recordings from 7,183 stereo-EEG (SEEG) contacts across 64 focal epilepsy patients, we extracted 260 features of local and network dynamics. Singular value decomposition revealed that ten eigenfeatures were sufficient to identify brain regions where seizures emerged (SZ). These eigenfeatures were used to train two unsupervised classifiers, which converged on a consensus model for assessing EpiNet-saliency using only the first and second eigenfeatures, reducing the dimensionality to 0.77% of the original feature set. The resulting model was tested on 7[ndash]9 hours of sleep-SEEG recordings from three patients in an independent cohort and validated using tensor component analysis. Among the 64 training subjects, SZ-classification accuracy correlated with individual position in the eigenfeature space (r2=0.5), suggesting that variability in EpiNet-saliency contributes to differences in classification performance. The model was tested on three subjects from an independent cohort, where it successfully captured time-varying EpiNet-saliency in sleep-SEEG, corroborating clinical assessments and achieving peak classification accuracies of 0.63, 0.85, and 0.94, respectively. A separate tensor component analysis verified the spatiotemporal characteristics of model-predicted EpiNet-saliency in the sleep-SEEG. These results provide compelling evidence supporting our hypothesis of a low-dimensional representation of epileptogenicity, which appears to be independent of brain state or neuropathological origin. This innovative approach enables straightforward interpretation of neuropathology and facilitates the integration of additional biomarkers and study of large cohorts, owing to its minimal computational footprint. These advancements significantly enhance the feasibility of a framework for unified epilepsy biomarkers.

Similar Papers

biorxiv
Mon Apr 21 2025
Flexible value coding in the mesolimbic dopamine system depending on internal water and sodium balance
Homeostatic imbalances elicit strong cravings, such as thirst and salt appetite, to restore equilibrium. Although midbrain dopaminergic neurons are known to encode the value of foods, their nutritional state-dependency remains unknown. Here, we show that the activity of the dopaminergic mesolimbic pathway flexibly expresses the positive and negative values of water and salt depending on the intern...
Ozawa, T.
•
Nakagawa, I.
•
Uchida, Y.
•
Abe, M.
...•
Hikida, T.
biorxiv
Mon Apr 21 2025
Breaking the Link: Neural Dissociation of Attention and Working Memory through Inhibitory Control
Attention and working memory (WM) encoding have traditionally been considered inseparable processes with shared neural mechanisms. Here, using an innovative experimental design and a multimodal approach, we provide the first direct neural evidence that attention and WM encoding are dissociable. Functional MRI identifies the supramarginal gyrus (SMG) as the key region enabling this dissociation, wh...
Liu, Y.
•
Fu, Y.
•
Tang, E.
•
Wu, H.
...•
Qin, P.
biorxiv
Mon Apr 21 2025
Feature-dependent decorrelation of sound representations across the auditory pathway
Early studies on orientation selectivity in the visual cortex have suggested that sensory systems generate new feature representations at specific processing stages. Many observations challenge this view, but in the absence of systematic, multistage measurements, the logic of how feature tuning emerges remains elusive. Here, using a generic approach based on representational similarity analysis wi...
Gosselin, E.
•
Bagur, S.
•
Jamali, S.
•
Puel, J.-L.
...•
Bathellier, B.
biorxiv
Mon Apr 21 2025
Plasticity of interhemispheric motor cortex connectivity induced by brain state-dependent cortico-cortical paired-associative stimulation
Transcallosal connectivity between the hand areas of the two primary motor cortices (M1) is important for coordination of unimanual and bimanual hand motor function. Effective connectivity of this M1-M1 pathway can be tested in the form of short-interval interhemispheric inhibition (SIHI) using dual-coil transcranial magnetic stimulation (TMS). Recently, we and others have demonstrated that the ph...
Ermolova, M.
•
Kozak, G.
•
Belardinelli, P.
•
Ziemann, U.
biorxiv
Mon Apr 21 2025
Inharmonicity enhances brain signals of attentional capture and auditory stream segregation
Harmonicity is an important feature for auditory perception as it influences pitch processing, memory and hearing in noisy environments. However, the neural substrates of processing harmonic and inharmonic sounds remain unclear. Here, we systematically manipulated the harmonicity of synthetic sounds by introducing random jittering to the frequencies above the fundamental. Using electroencephalogra...
Basinski, K.
•
Celma-Miralles, A.
•
Quiroga-Martinez, D. R.
•
Vuust, P.
biorxiv
Mon Apr 21 2025
Opponent regulation of striatal output pathways by dopamine and serotonin
Classic theories propose opponent functions for striatal dopamine (DA) and serotonin (5-hydroxytryptamine; 5HT), with DA promoting approach and 5HT promoting patience or avoidance. How these neuromodulators regulate downstream circuits to achieve such antagonistic effects remains mysterious. Here, we mapped striatal 5HT receptor expression and recorded from genetically-identified striatal neurons ...
Pinto, D. F. C.
•
Guo, M. Y.
•
Morishita, W.
•
Pomrenze, M. B.
...•
Malenka, R. C.
biorxiv
Mon Apr 21 2025
All spectral frequencies of neural activity reveal semantic representation in the human anterior ventral temporal cortex
Intracranial electrophysiology offers a unique insight into the nature of information representation in the brain - it can be used to disentangle information encoded in local neuronal activity (gamma and high gamma frequencies) from information encoded via long-range interactions (lower frequencies). We used regularised logistic regression to decode animacy from time-frequency power and phase extr...
Frisby, S. L.
•
Halai, A. D.
•
Cox, C. R.
•
Clarke, A.
...•
Lambon Ralph, M. A.
biorxiv
Mon Apr 21 2025
Stroke Shifts Brain Dynamics Toward Criticality: Evidence from Intrinsic Neural Timescales
Stroke induces widespread disruptions to brain function, extending beyond focal lesions to alter the multiscale temporal dynamics that govern neural processing. These dynamics-operating across milliseconds to months-form a hierarchical architecture essential for communication, integration, and adaptive plasticity. While stroke is known to impair local neural activity, its effects on this temporal ...
Wu, K.
•
Jelfs, B.
•
Fang, Q.
•
Gollo, L. L.
biorxiv
Mon Apr 21 2025
Causal Contributions of Left Inferior and Medial Frontal Cortex to Semantic and Executive Control
Semantic control guides the targeted and context-based retrieval from semantic memory. The overlap with and dissociation from domain-general executive control in the frontal lobe remains contentious. Here, we used transcranial magnetic stimulation (TMS) to probe the functional relevance of the left inferior frontal gyrus (IFG) and pre-supplementary motor area (pre-SMA) for semantic and executive c...
Martin, S.
•
Ferrante, M.
•
Bruera, A.
•
Hartwigsen, G.
biorxiv
Mon Apr 21 2025
Speech markers of psychedelic-induced psychological change
5-MeO-DMT, a potent, short-acting psychedelic, induces profound shifts in cognition, affect, and self-awareness. Because language explicitly expresses these domains and voice implicitly conveys them, both may serve as potential \'biomarkers\' of behavioural change. This study introduces a novel framework for analysing baseline language and vocal features, pre- to post-psychedelic changes, assessin...
Kuc, J.
•
McAlpine, R. G.
•
Sellers, A.
•
Blackburne, G.
...•
Skipper, J. I.