In this paper we present a multi-scale approach for cardiac modeling. Based on the histology of cardiac tissue we created a geometrical model at a cellular scale to compute the effective conductivity of a piece of cardiac tissue. In turn, the conductivity values obtained from this cellular scale model were used in a whole heart model in which we simulated regional, subendocardial ischemia. Histological changes at a cellular level led to changes in the effective conductivity tensor of the tissue, which in turn resulted in changes in the epicardial potential patterns during the ST-interval. Two effects were studied using this multi-scale approach: (1) the influence of a dynamically growing ischemic region on the epicardial potentials, and (2) the influence of a dynamically changing conductivity in the ischemic zone due to changes in the underlying pathology. One specific finding was the presence of epicardial patterns consisting of a central elevation and two opposite depressions at the edges of the ischemic zone which rotated as the ischemia became more transmural. In addition, the epicardial potentials decreased in magnitude with the duration of the ischemia due to changes in the effective conductivity of the ischemic tissue predicted by the cellular level model.