Low dimensional modeling of localized neural and hemodynamic response with habituation from multimodal brain measurements

Multimodality functional brain imaging has been gaining importance due to the complementary nature of different imaging modalities. Dynamical systems based models of neural activity and local hemodynamics may offer enhanced spatiotemporal resolution and insight into physiological signals and mechanisms. One major tool for studying brain function is to provoke local "evoked responses" by repeated application of a stimulus. Under such stimulation neuronal evoked responses show complex adaptation to stimuli as the repetition frequency increases beyond 2 Hz. These adaptation effects are also reflected in the resulting hemodynamic responses. The hemodynamic responses also change as a function of overall duration of a repeated string of stimuli. To understand the relationship between the adaptive neural activity and hemodynamics we propose a modeling framework that uses a simplified version of an existing neuron model with a control structure enabling adaptation prediction, and a neurovascular model that captures the relation between neuronal and hemodynamic evoked responses under different stimulus patterns. We report on the accuracy with which these models capture the neural and hemodynamic responses across a range of variation in stimulus patterns which includes both frequency and duration variation.