The Princess Margaret “Transformers”

Here we see how a pathologist (left) and a trained computer known as “ brAIn ” (right) visualize biopsy material from different brain tumours (e.g. glioblastoma). The brown colour highlights the computer-detected tumour within the patient’s tissue. The computer compares the patterns of this abnormal tissue to thousands and millions of previously seen examples and communicates its interpretation to the physician using prediction scores. 13 THE TRANSFORMERS • TECHNOLOGY ISSUE ACCELERATING CARE WITH ARTIFICIAL INTELLIGENCE THE brAIn PROJECT: ARTIFICIAL INTELLIGENCE AND DEEP LEARNING IN PATHOLOGY As a Neuropathologist and Scientist at The Princess Margaret, Dr. Phedias Diamandis specializes in diagnosing brain cancer. “Many critical decisions surrounding a patient’s treatment plan are based on their pathology reports. Accuracy is critical. I help provide some of the answers in hopes of the best possible outcome,” says Dr. Diamandis. Dr. Diamandis’ quest for precision led him to explore the benefits of artificial intelligence in pathology. “When I was training, the technology of scanning glass slides into digital images was in its infancy,” he explains. Today’s technology allows for the highest quality resolution. “Our research uses computers to help us precisely measure thousands of informative metrics, such as the diameter of tumour nuclei and the proportion of tumour cells dividing. This data, which we never had access to before, may help us refine our predictions of how individual tumours behave.” Over the past year, Dr. Diamandis’ research team developed an artificial intelligence tool they call “brAIn.” It uses cutting-edge technology called deep learning to automate diagnostics. “This technology could help transform routine pathology from qualitative art to quantitative science, and provide patients with more accurate diagnoses in shorter time frames.” This technology can be shared over the internet, too. Dr. Diamandis explains, “We hope to make brAIn a cost-effective and widely accessible diagnostic aid for remote cancer centres around the world.” Dr. Diamandis does not see artificial intelligence putting physicians out of work anytime soon. “Computers and humans excel at different aspects of image analysis, creating a dynamic duo for diagnostics. I see brAIn augmenting the services provided by pathologists and, in the end, patients will benefit the most.” WHAT IS THAT? PATHOLOGIST VIEW brAIn VIEW brAIn : Biological rendering through Artificial Intelligence and neural networks THE brAIn PROJECT