This is why +250 million mammograms were conducted worldwide last year.
But only 3 out of 100 screened women actually have suspicious findings and need to be recalled.
What if a machine learning model could rule out a large part of the obviously normal exams?
This would allow radiologists to focus on the remaining cases that have a higher likelihood of being suspicious.
Vara rules out normal mammograms with very high confidence to make sure human experts can focus on what really matters for the patients.