The gold standard of current treatment for atrial fibrillation is radiofrequency ablation (RFA). Single RFA procedures have low long-term, single-procedure success rates, which can be attributed to factors including inability to measure and visualize lesion depth in real time and incomplete knowledge of how atrial fibrillation manifests and persists. One way to address this problem is to develop a heart model that accurately fits lesion dimensions and depth using OCT to extract structural information. Twenty-three lesions of varying transmurality in left and right swine atrial tissue have been imaged with a Thorlabs OCT system with 6.5-micron axial resolution and a custom Ultra High Resolution system with 2.5-micron axial resolution. The boundaries of the ablation lesions were identified by the appearance of the birefringence artifact to identify areas of un-ablated tissue, as well as by changes to depth penetration and structural features, including decreased contrast between the endocardium and myocardium and disappearance of collagen fibers within the ablation lesion. Using these features, the lateral positions of the lesion boundaries were identified. An algorithm that fit ellipses to the lesion contours modeled the ablation geometry in depth. Lesion dimensions and shape were confirmed by comparison with trichrome histological processing. Finite-element models were fitted with these parameters and electrophysiological simulations were run with the Continuity 6 package. Next steps include correlating lesion geometry to conduction velocity, and including further tissue complexity such as varying tissue composition and fiber orientation. Additional models of linear lesions with gaps and adjacent lesions created with non-perpendicular contact will be created. This work will provide insight into how lesion geometry, tissue composition, and fiber organization impact electrophysiological propagation.