In the past decade, the field of deep learning has grown immensely and has shown significant success in automated analysis of complex data. In healthcare, deep neural networks have led to some bespoke clinical applications which learn directly from medical records and imaging data. This has led to some exciting applications which can potentially assist clinicians in their day-to-day workflow, and push the boundaries of clinical research. In this talk, I will discuss how deep neural networks have been leveraged at Sunnybrook to analyse and visualize microscopic images of breast cancer tissue. I will also describe the pros and cons of deep neural networks when adopted in healthcare, and propose some steps we can take to limit the need for large annotated datasets.

Brief Bio:

Dr. Shazia Akbar is a postdoctoral fellow at Sunnybrook Research Institute, Toronto, and Medical Biophysics, University of Toronto. Her specialty lies in the field of machine learning and medical image analysis, and currently she is investigating automated methods of predicting risk of breast cancer recurrence in digital pathology using deep learning. In 2015, Shazia completed her PhD at the University of Dundee, U.K., in collaboration with Ninewells Hospital, U.K. Shazia is also currently an affiliate of the Vector Institute in Toronto, Canada, and runs a machine learning journal club at Sunnybrook Health Sciences Centre.

Goldberg Computer Science Building, Dalhousie University

Date(s) - 07/01/2019
1:30 pm - 3:00 pm