Machine learning program classifies subtypes of lung cancer on par with pathologists

A new machine learning algorithm can help distinguish among the different subtypes of the most common form of lung cancer, which could lead to more accurate and faster diagnoses. Right now, trained pathologists look over slides of cancerous lung tissue in order to make the classifications, but the process can be time-consuming and highly subjective. Researchers at Dartmouth University’s cancer center tested the machine learning program against the diagnoses of three pathologists. Of the more than 140 slides that were tested, the program agreed with the pathologists’ conclusions nearly 70 percent of the time. But the researchers note that the data were from only one cancer center, and the technology needs more study before it reaches the clinic.