Analysing Microscopy Images with Deep Learning
Microscopy imaging is a rich source of biological information. It is widely used in medical practice (e.g. to diagnose breast cancer), in academia to study cells and industry for selecting suitable drug targets. Therefore, improving the analysis of microscopy images would result in a significant reduction of associated costs. One of the first steps in this analysis is identifying nuclei of cells. Currently this step is done either manually consuming thousands of man-hours or semi-automatically using algorithms that have to be adjusted for different types of experiments. Impressive generalisation capacity of deep neural networks holds a promise to empower fully automated systems for nuclei identification. For more than a year a team of researchers from the University of Tartu has worked closely with PerkinElmer – one of the leading companies in the area of microscopy, to bring the power of deep learning models into the cell analysis. In this talk will discuss most prominent results of this collaboration.