DeepButton: Automated image segmentation for challenging digital pathology data

by Yuchun Ding

16:00 (40 min) in USB 2.022

A major challenge in quantitative image analysis in digital pathology is the development of automated image segmentation methods. The existing software packages, such as CellProfiler and ilastik, have been commonly used by pathologists to perform quantitative analysis of histology data but these often failed on images stained with Hematoxylin and Eosin (H\&E) and immunohistochemistry (IHC) when segmenting various types of cell nuclei on complex tissue structures. In recent years, a rapid growing area in computer science, deep learning, is becoming a solution to overcome this issue. However, an important usability limitation with current deep learning software is that it is provided in the form of toolboxes or libraries that require extensive computer knowledge to install and implement. We therefore developed DeepButton to overcome some of these limitations. The software uses pre-trained deep learning models to provide easy-to-use solutions for automating the segmentation of difficult images.