The re-endothelialization of an acellular lung scaffold is evaluated through the analysis of histological images, which is an unscalable process due to its time-consuming, work-intensive, and subjective nature. Deep learning methods present a means to accurately automate medical image analysis and in particular, semantic segmentation can be used to evaluate re-endothelialized lungs. In this thesis, we investigated the ability of semantic segmentation to determine key performance metrics in re-endothelialized mouse lungs. The cell seeding coverage, which quantifies the distribution of seeded cells in the lung scaffold, was accurately predicted by our implementation of semantic segmentation. Ruptured and dilated vessels, which deform the lung and signify problems with the recellularization procedure, were segmented with potential for improvement. Through patch-based learning, data augmentation, and transfer learning techniques, we significantly ameliorated model performance. We find that our work provides a proof-of-concept for the standard use of semantic segmentation to analyze recellularized lung images.