AI in pathology: paving the way forward
In the past years, the game has changed for computational pathology. Thomas Fuchs, Founder and Chief Scientist at Paige: “For a long time no one really cared, but since the first digital high resolution images for clinical use, it really took off and is now exploding. The rise of image-based technologies and deep learning open the door to AI in pathology. We can now develop possible models and systems that are accurate (and predictive) enough to use for clinical tasks.”
“Since the first digitized high resolution images for clinical use, computational pathology is really taking off. The rise of image-based technologies and deep learning open the door to AI in pathology. We are now able to develop possible models and systems that are accurate (and predictive) enough to use for clinical tasks.”
Why Bigpicture is essential
A platform such as Bigpicture is important for the development of AI in pathology. With 45 partners in both the public and private sector, and with all kinds of expertise involved, this is the moment where we can get the work done that will push pathology into the next phase of science. Thomas was invited as a keynote speaker during Bigpicture’s Annual Meeting in February this year. Thomas: “The goal of Bigpicture, to push computational pathology to the next level, is a necessity. There are not enough pathologists, the amount of cancer patients goes up, the complexity of the work is increasing, and thus is the workload. AI can really make a difference here.”
Luckily, there’s a large choice in barriers and drugs that can help patients. But also in this case, the field is changing. For instance the increasing demand of precision medicine (each individual patient needs the right medicine). Thomas: “AI can have a vital role in accurately selecting medicine for patients.”
“Bigpicture brings so many stakeholders and different expertise together. It provides the resources for a community that operates in an industry that is still fractured. Plus, there really is a lack of public datasets due to privacy restrictions and procedures. To get valuable data available for (AI) researchers will have an enormous impact on computational pathology in the years to come. This is obviously very beneficial for the research community.”
It all starts with digitization
We are still in the early days of computational pathology, with the digitization of whole slide images as one of its current main challenges. Thomas: “The Netherlands is one of the leading countries in digitizing slides, but in many countries hospitals are still working traditionally with a microscope and are far from digitizing their slides. In many cases, pathology hasn’t changed in the past 150 years! As digitization is the basis of computational pathology, it is absolutely critical that this needs to change. The Bigpicture initiative helps in forcing a movement in the industry.”
Pathology in the coming years
Thomas suspects that in 5 years’ time 25% of the hospitals will be digitized (in comparison to a current ~5%), and 90% in the coming 10 years. The next decade will show significant changes in the world of pathology. Thomas: “In my lab we have already proven that AI algorithms can easily find cancers. I expect that AI will be part of the workflow for all big cancers within the next 5 years. It will be a huge improvement and makes the work for a pathologist much easier. Imagine having all cases you need in a few clicks, without having to wait, and the ability to immediately find cancer thanks to the use of algorithm tools.”
Dr Thomas J. Fuchs is the Founder and Chief Scientist at Paige. He is a pioneer in the groundbreaking field of computational pathology whose inventions led to the first FDA-approved AI-based pathology product, Paige Prostate. Thomas was named one of the Top 100 AI Leaders in Drug Discovery and Advanced Healthcare in 2019. He is also the Dean of Artificial Intelligence and Human Health and Co-Director of the Hasso Plattner Institute for Digital Health at the Icahn School of Medicine at Mount Sinai.