Higher Education Teaching Material on Machine Learning in the Domain of Digital Pathology

Machine learning (ML) and especially deep learning (DL) are doubtless one of the key technologies of the last couple of years and future decades. Learning its theoretical concepts alone is a big challenge as it requires a strong background in mathematics and computer science. Once students and researchers have built up on the core concepts and recent algorithms using toy data sets and want to move forward tackling real-world application scenarios, they are confronted with additional problems, specific to the application domain. Digital pathology is one of those domains that has many of these additional issues, such as: New file formats, large images, and the amount of medical expertise needed to capture the underlying problem is relatively large. In our project deep-teaching.org we provide teaching materials to introduce computer science students to the use of ML for digital pathology. As example, we present the application of pathologic N-stage (pN-stage) classification by dividing it into sub-tasks and propose - step by step - how different machine and deep learning, as well as computer vision techniques, can be combined to contribute to the main objective.