Open Source, Open Minds

By March of 2020, the world was acutely aware that with COVID-19, we had a significant problem on our hands. Not yet fully understanding the impact of the illness, machine learning and computer vision researchers at the University of Waterloo and DarwinAI were looking for a way to leverage automation, computational analysis and AI/ML to speed the identification of COVID infection - helping to clear the bottleneck in triage and diagnosis. They developed an open source model that could quickly scan patient chest images and detect whether or not patients had COVID, pneumonia, or healthy lungs. They didn’t have a platform to run it on, though, nor any ability to deploy it for medical practitioners to access it.

Enter the ChRIS Project - a collaboration between Boston University, Red Hat and the Boston Children’s Hospital. ChRIS is an open source medical analysis platform built on top of Red Hat’s OpenShift platform, and the ChRIS team was at the same time actively looking for other open source projects to collaborate with to help fight COVID. The COVID-Net team provided the model, and ChRIS provided the platform including a backend built for processing medical imaging data and all the plumbing needed to stand up a clinician-focused UI. Alongside Red Hat volunteers, an open source COVID-fighting collaboration was born. In less than a week, the COVID-Net model was packaged as a plugin and running on the ChRIS platform, and in just a couple of months Red Hat volunteers designed a UI front end that the COVID-Net team implemented on top of ChRIS, and the team is looking to start testing this UI with clinicians in early 2021.  

Hear more about the ChRIS Project on this episode of Open Source, Open Minds.

Speakers:

Rudolph Pienaar, Technical Director of the Fetal-Neonatal Neuroimaging and Development Science Center (FNNDSC), Boston Children’s Hospital