ABSTRACT
This chapter introduces deep learning at a very high level, motivating the more detailed examination of deep learning for organ-at-risk segmentation in this part of the book. An overview of the historical background to deep learning is provided, giving the reader some insight into its early origins and at the same time explaining why the approach has recently become more popular despite the original concept dating back over 50 years. A brief discussion is explored as to how deep learning-based auto-contouring differs from atlas-based or model-based segmentation, and why it has proved more powerful than these alternatives. The chapter finishes with an overview of the other chapters on aspects of deep learning for auto-segmentation of organs-at-risk in this part of the book.
