Great interest in digital pathology

Around 60 people attended the SciLifeLab workshop in Uppsala. Here, Peter Bankhead from Belfast is lecturing on a digital pathology software application.

Which digital tools can help pathologists with the analysis of samples? This was the theme of a workshop on 9 November, for both computer scientists and pathologists, at Uppsala University’s Department of Information Technology.

Professor Carolina Wählby.
Photo: Mikael Wallerstedt

“We had counted on 20 visitors, but about 60 turned up. It was wonderful that so many came from the medicine side,” says Carolina Wählby, Professor at the Department of Information Technology and Director of SciLifeLab’s BioImage Informatics Facility.

During the workshop an open-source software solution used at the Facility, was presented.

“We usually work with free software, to offer flexibility and not depend on licences. This is the first really good open-source digital pathology tool, and the open-source code allows us to build it up with new features,” Wählby says.

At the workshop there was also a lecture by Jeroen van der Laak of Radboud University Medical Center in Nijmegen (the Netherlands), which is world-leading in artificial intelligence and pathology. On the following day, he served as the opponent at Sajith K. Sadanandan’s  PhD disputation (public thesis defence). The lecture therefore attracted many people outside the Department of Information Technology as well.

“There’s an incredible lot happening in digital pathology, both on the research side and in companies that see the potential of the new tools,” Wählby says.

Twenty years have passed since X-ray images went from paper to digital. Now the same change is taking place in pathology. Today’s pathologists can scan their tissue samples and view them on a computer screen rather than leaning over a microscope.

“Instead of posting samples on glass slides to colleagues and external experts, we can email hyperlinks. Switching from microscope to monitor gives us opportunities for using digital tools.”

Deep learning enables the computer to be trained to recognise structures and patterns. The images are so large that they cannot be seen in their entirety in high resolution. The computer can process all the information, find suspect structures and guide the pathologist, who is then the one who makes decisions.

“One of the participants compared pathologists to pilots – who are responsible for human life too – and how they use computer support. In healthcare as well, help in making decisions is needed. I think many people see this not as a threat, but as an opportunity,” Wählby says.

“With digital tools you can eliminate a lot of the drudgery and focus on the important questions.”

Find out more

The BioImage Informatics Facility

5 december 2017