AI to optimize breast cancer diagnostics


Thanks to a SEK 20 million grant, a research team is going to develop an AI system to improve diagnostic imaging for breast cancer. The pathology research is headed by Johan Hartman at Karolinska University Hospital.

"We're building an AI system that will be able to independently locate and diagnose cancer in the microscopy images we work with. The aim is that the system will serve as a basis of decision for radiologists and pathologists when diagnosing breast cancer," says Johan Hartman, specialist in Clinical Pathology and Cytology and head of the research team.

Microscopy images are currently analyzed digitally, but many diagnoses fall into a gray zone because certain tumors are not assessed to be aggressive, yet are not fully benign either. The AI system will reduce the number or patients in this gray zone.

"Today the image assessments vary widely because we humans assess them differently. But AI ought to be able to find information that we can't spot in the images. This will help us a lot in our work, and more patients will receive a clearer diagnosis and thus the right treatment for them," Hartman says.