Scientists at New York University School of Medicine have successfully used a sophisticated computer program to distinguish between different types of lung cancer and identify the mutated genes behind them.
The developers say the technology could open the door to faster diagnosis of other types of cancer, including malignant mesothelioma.
Malignant Mesothelioma, the Asbestos Cancer
Malignant pleural mesothelioma is a form of lung cancer caused by asbestos exposure.
Instead of starting in the lung tissue like other types of lung cancer, it grows on the membrane that surrounds the lungs and can then spread to the lungs and other organs. There is no cure and the prognosis for mesothelioma is usually poor.
As with most types of cancer, earlier diagnosis of mesothelioma can improve the odds of survival by making earlier intervention possible. Experts hope that the newly developed artificial intelligence or “machine learning” program, detailed online in Nature Medicine this week, can help with this often-difficult task.
Artificial Intelligence and Mesothelioma Diagnosis
Using data from The Cancer Genome Atlas, the researchers fed images of slides from people with and without lung cancer into an artificial intelligence (AI) program developed by Google and asked it to identify the cancer cases and classify them by cancer type.
According to the NYU team, the program was able to distinguish between adenocarcinoma and squamous cell carcinoma with 97 percent accuracy — about as good as an experienced pathologist.
In addition, the researchers also “taught” the program to identify six of the genes linked to lung cancer (two of which have also been linked to pleural mesothelioma) and detect any genetic abnormalities. Accuracy for that task ranged from about 73 to 86 percent, depending on the gene.
What Does it Mean for Mesothelioma Patients?
The results of this study could be good news for future patients with suspected pleural mesothelioma and other cancers that normally depend heavily on skilled pathological analysis.
In its earliest, most-treatable stages, mesothelioma often causes vague symptoms like coughing and chest pain, both of which can also be caused by other, less serious issues.
While there is technology in existence that can help pathologists distinguish certain kinds of cancer cells and identify mutated genes, these often take weeks to return results while the new AI program did it in minutes.
“Delaying the start of cancer treatment is never good,” says lead investigator Aristotelis Tsirigos, PhD, associate professor in the Department of Pathology at NYU Langone’s Perlmutter Cancer Center. “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.”
Targeted cancer therapies, based on a patient’s unique genetic profile, have the potential to produce better treatment outcomes.
NYU Langone Health says the team plans to continue to “teach” the program to recognize genetic mutations in multiple types of cancer before it seeks government approval to use the technology with patients.
Coudray, N, et al, “Classification and mutation prediction from non–small cell lung cancer histopathology images using deep learning”, September 17, 2018, Nature Medicine
“Artificial Intelligence Tool Accurately Identifies Cancer Type & Genetic Changes in Each Patient’s Lung Tumor”, September 17, 2018, News Release from NYU Langone Health/NYU School of Medicine