Cortical depth-dependent modeling of visual hemodynamic responses

•We develop a physiologically based 3D hemodynamic model to capture 3D blood flows.•The model accurately predicts empirical visual BOLD-fMRI responses versus cortical depth.•The model provides better predictions and is more parsimonious than existing models.•The model allows better physiological understanding of the depth-dependence of visual BOLD responses. A physiologically based three-dimensional (3D) hemodynamic model is developed to predict the experimentally observed blood oxygen level dependent (BOLD) responses versus the cortical depth induced by visual stimuli. Prior 2D approximations are relaxed in order to analyze 3D blood flow dynamics as a function of cortical depth. Comparison of the predictions with experimental data for evoked stimuli demonstrates that the full 3D model performs at least as well as previous approaches while remaining parsimonious. In particular, the 3D model requires significantly fewer assumptions and model parameters than previous models such that there is no longer need to define depth-specific parameter values for spatial spreading, peak amplitude, and hemodynamic velocity.