AI identifies lung cancer type with 97% accuracy

An artificial intelligence (AI) algorithm created by NYU School of Medicine researchers distinguished between two forms of lung cancer with 97 percent accuracy, according to a new study published in Nature Medicine.

Adenocarcinoma and squamous cell carcinoma are two cancer types pathologists struggle to identify from one another without testing, according to the study. The AI method could detect if an abnormal version of six genes associated with lung cancer were existent in cells, all without gene testing.

Targeted therapies that only work against cancer cells with specific mutations have gained traction. For example, nearly 20 percent of patients with adenocarcinoma have a mutation that can be treated with approved drugs, according to the study. However, genetic testing verifying those genes can take weeks.

"Delaying the start of cancer treatment is never good," said senior author Aristotelis Tsirigos, PhD, 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."

Tsirigos and colleagues trained Google’s Inception v3 deep convolutional neural network to study pathology slide images from the Cancer Genome Atlas database.

Results showed 45 of the 54 images misclassified by at least one pathologist were correctly identified according to cancer type by the AI algorithm. Additionally, half of the small percentage of tumor images misclassified by the AI platform were also wrongly identified by pathologists.

According to the authors, their results demonstrate AI’s ability to lend a second opinion to complicated diagnostic cases.

"In our study, we were excited to improve on pathologist-level accuracies, and to show that AI can discover previously unknown patterns in the visible features of cancer cells and the tissues around them," said a corresponding author on the study Narges Razavian, PhD. "The synergy between data and computational power is creating unprecedented opportunities to improve both the practice and the science of medicine."