Self-Supervised Learning for Histopathology with Jean-Baptiste Schiratti from Owkin

In this episode, I sit down with Jean-Baptiste Schiratti, Medical Imaging Group Lead and Lead Research Scientist at Owkin, to discuss the application of self-supervised learning in drug development and diagnostics. Owkin is a groundbreaking AI biotechnology company revolutionizing the field of medical research and treatment. It aims to bridge the gap between complex biological understanding and the development of innovative treatments. In our conversation, we discuss his background, Owkin's mission, and the importance of AI in healthcare. We delve into self-supervised learning, its benefits, and its application in pathology. Gain insights into the significance of data diversity and computational resources in training self-supervised models and the development of multimodal foundation models. He also shares the impact Owkin aims to achieve in the coming years and the next hurdle for self-supervised learning.

Key Points:

• Introducing Jean-Baptiste Schiratti, his background, and path to Owkin.
• Details about Owkin, its mission, and why its work is significant.
• The application of self-supervised learning in drug development and diagnostics.
• Examples of the different applications of self-supervised learning.
• Discover the process behind training self-supervised models for pathology.
• Explore the various benefits of using self-supervised learning.
• His approach for structuring the data used for self-supervised learning.
• Unpack the potential impact of self-supervised AI models on pathology.
• Gain insights into the next frontier of foundation model development.
• He shares his hopes for the future impact of Owkin.

Quotes:

“To be able to train efficiently, computer vision backbones, you actually need to have a lot of compute and that can be very costly.” — Jean-Baptiste Schiratti

“There are some models that are indeed particular to specific types of tissue or specific sub-types of cancers and also the models can have different architectures and different sizes, they come in different flavors.” — Jean-Baptiste Schiratti

“The more diverse the [training] data is, the better.” — Jean-Baptiste Schiratti

“I’m convinced that the foundation models will play a very important role in digital pathology and I think this is already happening.” — Jean-Baptiste Schiratti

Links:

Jean-Baptiste Schiratti on LinkedIn

Jean-Baptiste Schiratti on X

Owkin

Phikon

Resources for Computer Vision Teams:

LinkedIn – Connect with Heather.

Computer Vision Insights Newsletter – A biweekly newsletter to help bring the latest machine learning and computer vision research to applications in people and planetary health.

Computer Vision Strategy Session – Not sure how to advance your computer vision project? Get unstuck with a clear set of next steps. Schedule a 1 hour strategy session now to advance your project.

Foundation Model Assessment – Foundation models are popping up everywhere – do you need one for your proprietary image dataset? Get a clear perspective on whether you can benefit from a domain-specific foundation model.