Peter Bankhead Research Group

Section: Translational Epidemiology

Research in a Nutshell

Image

We are a computational research group who develop methods and software to analyse biomedical images. We are especially enthusiastic about coming up with practical tools that solve tricky image analysis problems in new and intuitive ways. We believe in the importance of open science; we play our part by creating open software used by thousands of researchers worldwide.

Our work has applications across basic biology and digital pathology. A major challenge when working with pathology images is that they are both big and complex. A typical scan of a tissue sample can be 50 GB in size and contain millions of cells. These images encode a wealth of information that could further our understanding of cancer and other diseases. However, it remains extremely difficult to extract this information to answer precise scientific questions.

We tackle these challenges using a variety of techniques, including multiscale image processing, machine learning, and spatial analysis. Through collaborative projects, we develop algorithms for specific applications (e.g., assessing biomarkers in breast cancer, quantifying myelin in brain tissue for multiple sclerosis research) while always thinking about how to generalize these algorithms to different datasets and applications. We challenge ourselves to move beyond developing ‘proof-of-concept’ methods into providing tools that are as useful as possible to the wider scientific community.

To achieve this, we also develop and support QuPath. QuPath is user-friendly, open-source software that provides us with a platform both to develop and share algorithms. Since first created by Peter when he was a postdoc, QuPath has been downloaded over 250,000 times, cited in over 1,500 publications, and has a growing user community across academia and industry. With support from the Chan Zuckerberg Initiative (Essential Open Source Software for Science) and the Wellcome Trust (Technology Development Grant), we are expanding QuPath to solve an even wider range of biomedical and microscopy image analysis problems.

Our work involves collaborating and communicating. We’re comfortable with code and pixels, but we want our methods to be understood by specialists in other areas. To this end, Peter has also written a popular open handbook, ‘Introduction to Bioimage Analysis’, that aims to make core image analysis concepts and techniques accessible and intuitive.

Links

QuPath (external)

"Introduction to Bioimage Analysis" textbook (external)

Image

People

Contact

p.bankhead@ed.ac.uk