Building an Automated Cellular Annotation Framework

Research in Leishmania and related parasites produces large amounts (up to the thousands) of microscopy images, which, in turn, require large amounts of time to classify and annotate. Not only does this detract the researchers from exploring new alternatives by robbing them of useful time, as it is also prone to inter-person variance. This lead us to the need for automatic or semi-automatic methods for image classification and annotation algorithms. In this book we guide the reader through the basic building blocks of a multi-layer computer vision / AI hybrid algorithm designed to automatically process confocal microscopy images in a consistent pipeline. In addition, a thorough validation of the algorithm is presented in a two-stage process. Firstly, by analysing each of its constituent components and secondly, by pitying it against a team of expert biomedical researchers to assess its real-world performance, as well as to reveal potential future improvements.