An open-source, MATLAB based annotation tool for virtual slides

Abstract

Image annotation is an important step in the development of automated analysis methods for digitized microscopy samples. Annotated areas (i.e. regions of interest) are used both during the training process and for evaluation of performance of an automated tool. For example supervised learning algorithms require a large number of training samples to adequately learn a model for a particular task. Current virtual slide collections on the other hand contains vast amounts of data and new tools are needed to perform image annotation in a virtual microscopy environment. The challenge is to extract areas of interest including labels from digital whole-slide samples in an efficient and easy-to-use manner.