The cerebrospinal fluid is a crucial element of the cerebral compartment ensuring the correct functioning of the brain. Its clearance and renewal is thus very important and has been closely related to the cerebral venous system, through the glymphatic system. It motivates the study of the blood dynamic in the cerebral venous network. Such network evolves complex geometries leading to fully three dimensional flows. In addition, the venous network is subject to inter-individual variability and then requires individual-specific processing. In this context we developed a numerical framework starting from MRI and phase-contrast MRI (PC-MRI) to simulations. Both MRI and PC-MRI are non-invasive imaging techniques providing, respectively, structural and physiological information. After this image processing task, we use a numerical model based on Navier-Stokes equations coupled with Windkessel boundary conditions to simulate the blood dynamic. These boundary conditions allow to encode truncated part of the network, but require to tune many parameters. To do so, we use flow rate measurements at specific location of the network contained in the PC-MRI acquisitions. The fine parameter tuning remain a challenging task but can be automatize using data assimilation methods based on Kalman inversion [1]. It however remains computationally expensive and can be speedup using surrogate modelling.