Study: Artificial intelligence determines lung cancer type with 97% accuracy

NEW YORK — Artificial intelligence is showing its value as a critical tool in the doctor’s office, particularly when it comes to accurately diagnosing serious ailments. Researchers at the New York University School of Medicine created a computer program that can analyze images of lung tumors in patients, specify cancer types, and identify altered genes that cause abnormal cell growth.

The researchers behind the latest breakthrough in medical artificial intelligence found their program could distinguish between adenocarcinoma and squamous cell carcinoma — two types of lung cancer that are particularly difficult to differentiate — with 97% accuracy.

The program also determined whether abnormal mutations of six separate genes linked to lung cancer were present in cells with an accuracy rate of 73% to 86%, depending on the gene. These genetic mutations commonly cause abnormal growth related to cancer and also change a cell’s shape and its interactions with other cells.

When treating cancer, physicians try to determine which genes are changed in each tumor. New targeted therapies operate on this principle, working only against cancer cells identified by specific mutations.

Current genetic testing methods can take weeks to return results, the researchers say.

“Delaying the start of cancer treatment is never good,” says senior study author Dr. Aristotelis Tsirigos, associate professor in the Department of Pathology at NYU Langone’s Perlmutter Cancer Center, in a statement. “Our study provides strong evidence that an AI approach will be able to instantly determine cancer subtype and mutational profile to get patients started on targeted therapies sooner.”

Dr. Tsirigos and his team trained a deep convolutional neural network, in this case Google’s Inception v3, to analyze slides culled from The Cancer Genome Atlas, a database of images of cancer diagnoses that have already been confirmed. This allowed the researchers to test their AI program in its ability to accurately and automatically classify normal and cancerous tissue.

The study was published in the journal Nature Medicine.

Leave a Reply

Your email address will not be published. Required fields are marked *