The goal of this study is to present a methodology to create 3 dimensional structures reconstructed from high resolution histological sections by utilizing convolutional neural networks to annotate the images, discard the folded and/or torn section with an unsupervised information-theory-based clustering algorithm and apply Homography to correct perspective of each image. The resulting models then can be used for 3 dimensional micromorphometry analysis to aid the decision making process of medical professionals.