A multi-scale computational model of cortical spreading depression and thalamocortical sensory gating
Biomedical Signal Processing and Control, cilt.128, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 128
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.bspc.2026.111244
- Dergi Adı: Biomedical Signal Processing and Control
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, EMBASE
- Anahtar Kelimeler: CSD, Filippov systems, Lyapunov stability, Parameter identifiability, PRCC sensitivity, Sensory gating, Singular perturbation
- Hacettepe Üniversitesi Adresli: Evet
Özet
Cortical spreading depression (CSD) is a slowly propagating wave of near-complete neuronal and glial depolarisation regarded as the electrophysiological substrate of migraine aura. We propose a multi-scale nonlinear dynamical model integrating fast thalamocortical paired-pulse dynamics with a slow CSD modulation field. The model is analysed rigorously: we establish existence and uniqueness of solutions via the Picard–Lindelöf theorem with explicit Lipschitz constants, prove Lyapunov stability of the resting equilibrium with quantitative basin estimates, perform Tikhonov–Fenichel singular perturbation reduction yielding the slow invariant manifold, and carry out Jacobian-based bifurcation analysis. The model is calibrated against 67 paired-pulse SEP recordings from 14 Wistar rats, of which nine yielded analysable datasets (11 recording configurations, 1244 sweeps); all inference is at the level of recordings and animals. Cohort-averaged results show proportional suppression of S1 (37.5%) and S2 (36.5%) during early post-CSD, preservation of G(50ms)=0.63±0.03, and S1 recovery from early to 10 min post-CSD that is consistent in direction across all eight animals (exact Wilcoxon p=0.0078, the attainable floor at n=8; Holm-adjusted p=0.031). A formal identifiability analysis (profile likelihood, collinearity index, parametric bootstrap) shows that the recovery rate μK and the effective inhibitory time constant are determined by these data whereas two further parameters are not, and a model comparison against a constant-gain null model and an equally parameterised descriptive model shows that a phase-dependent modulation is strongly supported and that the mechanistic formulation reduces held-out long-ISI prediction error by 60%. Global PRCC sensitivity analysis identifies μK as the parameter to which recovery duration is most sensitive; this is a model-derived hypothesis requiring targeted experimental test.