Researchers in Zurich are using supercomputing technology to wade through piles of medical data to help them diagnose prostate cancer.
They are feeding the IBM supercomputer Watson biopsy slides—sometimes as many as 50 per patient—that the computer analyzes for signs of tumors.
A pathologist takes around an hour and a half to analyze one patient, whereas Watson can do it in less than a second.
The researchers believe that the computer is currently more than 99 percent accurate.
“If the pathologist becomes faster by using such a system I think it will pay off,” said Dr. Peter Wild. “If I sit here one and a half hours looking at slides, screening all these slides, instead of just signing out the two or three positive ones, and taking into account that there may be a .1 error rate, percent error rate, this will pay off.”
The hospital is slowly feeding its entire biopsy archive, thousands of slides from the past 15 years, to the supercomputer, a process that could take years to complete.