Neuronal and hemodynamic source modeling of optogenetic BOLD signals

An approach to model the hidden neuronal and hemodynamic sources of the blood-oxygenation-level-dependent (BOLD) signal observed in optogenetic functional MRI experiments is presented. In these experiments, genetically modified light-sensitive neurons are directly stimulated by laser light. A proof-of-principle demonstration is provided by estimating hidden hemodynamic and neuronal states in the Buxton-Mandeville-Friston BOLD model applied to simulated and experimental optogenetic fMRI data of the mouse. The estimation procedure is based on the continuous unscented Kalman filter. In simulations, under a given model, unobserved BOLD source signals such as cerebral blood flow and deoxygenation could be estimated with good accuracy, although neuronal signal sources appeared lagged in time. In experimental data, this approach was able to describe positive and negative BOLD signals. Translated into human applications, it could have the potential to provide novel functional markers for imaging cerebrovascular accidents or altered vascularity of brain tumors in a more specific way than standard clinical MRI protocols.