Image-based computational fluid dynamics (CFD) has become a new capability for determining wall stresses of pulsatile flows. However, a computational platform that directly connects image information to pulsatile wall stresses is lacking. Prevailing methods rely on manual crafting of a hodgepodge of multidisciplinary software packages, which is usually laborious and error prone. We present a new technique to compute wall stresses in image-based pulsatile flows using the lattice Boltzmann method (LBM). The novelty includes: (1) a unique image processing to extract flow domain and local wall normality, (2) a seamless connection between image extraction and CFD, (3) an en-route calculation of strain-rate tensor, and (4) GPU acceleration (not included here). We first generalize the streaming operation in the LBM and then conduct an application study for laminar and turbulent pulsatile flows in an image-based pipe (Reynolds number: 10 to 5000). The computed pulsatile velocity and shear stress are in good agreement with Womersley solutions for laminar flows and concurrent laboratory measurements for turbulent flows. This technique is being used to study (1) the hemodynamic wall stresses in inner choroid endothelium, (2) the drag force in sand flows, and (3) effects of waste streams on ion exchange kinetics in porous media.