In the seminar "Machine Learning in Medicine - Theory & Practice" you will implement state-of-the-art supervised segmentation methods for medical image data, e.g., computed tomography (CT). After an introduction to the data we will create a common evaluation pipeline. Then, in small teams, we will study selected key publications of the most popular network architectures, such as U-Net, Densenet, and ResNet. You will implement and evaluate the algorithms on the medical CT data. The results of the teams will be compared and the influence of hyperparameter choices will be investigated. Note, there will be a strong focus on the practical implementation of the algorithms, requiring a sound programming knowledge in python.