Authors: Alicia Gonsalves¹; Ryszard Auksztulewicz¹; Michelle Moerel¹
¹Maastricht University
Background: Laminar fMRI offers a noninvasive view of cortical microcircuits, but interpreting these signals is challenging, as BOLD reflects hemodynamic rather than neuronal activity directly. This requires computational models that can disambiguate this relationship. Within Dynamic Causal Modelling (DCM), this is achieved in P-DCM by pairing a neuronal model with a hemodynamic model correcting for ascending vein effects in laminar BOLD. Yet, current work models the neuronal layers without interlaminar connectivity essential to cortical processing. We extend this framework by including interlaminar connections to capture cortical information flow.
Method: We validate a single-region P-DCM comprising a neuronal model (N=3 layers), neurovascular coupling, and a hemodynamic model (K=7 depths) (Fig. 1). Forward simulations establish face validity, and parameter recovery assesses identifiability of interlaminar connectivity (A), modulation (B), driving input (C), and neuronal parameters (σ, μ, λ).
Results: Simulations show plausible layer-specific dynamics and distinct BOLD depth profiles (Fig. 2). Following driving input, the middle (M) layer responded first, propagating activity to superficial (S) and deep (D) layers via A. Modulation increased responses globally. Parameter recovery from noise-free simulations showed partial identifiability (Fig. 3). C, A (M→S), A (S→D), and μ recovered well. B mirrored A but degraded: M→S recovered well; S→D and D→M showed reduced accuracy. σ and λ showed range-dependent recovery. D→M showed flat recovery for A and B.
Conclusion: These results show that the model captures core dynamics and clarifies identifiability boundaries. Poor recovery of D→M and λ likely reflects BOLD’s limited temporal resolution, inadequately constraining fast neuronal dynamics. Next steps involve empirical 7T fMRI validation and Bayesian fusion with iEEG. The model will be the foundation for a multi-region framework to study auditory microcircuit dynamics.
