Determining whether cancer has spread, or metastasized, from the primary site to nearby lymph nodes is a difficult, time-intensive task for pathologists. Google AI has developed a promising algorithm to evaluate lymph node biopsies in breast cancer patients, and it’s more accurate than humans in certain circumstances.
In new journal articles, Google AI says its deep-learning program called the LYmph Node Assistant, or LYNA, was able to distinguish between slides with or without metastatic cancer 99% of the time, and locate the suspicious sites, even when looking for extremely small metastases human pathologists might miss. LYNA also cut the slide review time from two minutes to one minute.
LYNA wasn’t perfect — it occasionally misidentified giant cells, germinal cancers, and bone marrow-derived white blood cells known as histiocytes — but managed to perform better than a pathologist. Human pathologists miss small metastases on individual slides as much as 62% of the time when under time constraints.
Google (GOOGL) researchers said the LYNA program would require more testing but was a promising step in creating AI technology to assist doctors and clinicians.
Google’s play in health care AI is big. In June, its Medical Brain team said it had created an AI system that could forecast mortality rates with 90% accuracy. Google’s Verily also has made inroads with determining a person’s risk of heart disease with retinal scans. A recent study found a computer was a more likely to identify dangerous skin lesions than dermatologists, and Japanese researchers found AI was promising in detecting early-stage colorectal cancer.