The Future of Radiology
Radiologists are increasingly using deep learning algorithms to identify diseases in medical scans of patients. But who is responsible if the programs make a mistake?
When Regina Barzilay had a routine mammogram in her early 40s, the image showed white patches in her breast tissue. This can be a sign of cancer or it can be completely harmless - even the best radiologists often have a hard time telling the difference. Barzilay's doctors were optimistic and said there was no need to worry immediately. "I already had cancer, but they didn't see it," says Barzilay afterwards.
Over the next two years, she underwent a second mammogram, an MRI, and a biopsy, all of which continued to provide unclear or even conflicting results. Finally, in 2014, the doctors diagnosed breast cancer; just getting there was incredibly frustrating for Barzilay. "How can you take three tests and get three different results?" she wondered.
The cancer is now cured. But Barzilay remained appalled that the treatment of many patients is delayed because the results of a mammogram cannot always be correctly interpreted. She really wanted to change that - and therefore made a decision that changed her career and her life forever …