Résumé : The aim of this work was to develop a simple and fast 3D MRI-based computer model of arrhythmia inducibility in porcine hearts with chronic infarct scar, and to further validate it using electrophysiology (EP) measures obtained in-vivo. The heart model was built from MRI scans (with voxel size smaller than 1mm3) and had fiber directions extracted from diffusion tensor DT-MRI. We used a macroscopic model that calculates the propagation of action potential (AP) after application of a train of stimuli, with location and timing replicating precisely the stimulation protocol used in the in-vivo EP study. Simulation results were performed for two infarct hearts: one with noninducible and the other with inducible ventricular tachycardia (VT), successfully predicting the study outcome like in the in-vivo cases; for the inducible heart, the average predicted VT cycle length was 273ms, compared to a recorded VT of approximately 250ms. We also generated synthetic fibers for each heart and found the associated helix angle whose transmural variation (in healthy zones) from endo- to epicardium gave the smallest difference (i.e., approx. 41°) when compared to the helix angle corresponding to fibers from DW-MRI. Mean differences between activation times computed using DT-MRI fibers and using synthetic fibers for the two hearts were 6 ms and 11 ms, respectively.