September 9, 2021

Artificial Intelligence Imaging Program Detects Lung Nodules That Are Likely to Become Malignant

By Ed Susman

VIRTUAL -- September 9, 2021 -- An artificial intelligence imaging program may be able to give doctors a heads up on what lung nodules are likely to become cancerous when compared with current detection methods, researchers reported at the 2021 virtual European Respiratory Society’s International Congress.

“A deep learning system for lung nodule detection from low-dose CT scans showed good performance at detecting malignant biopsy-confirmed lesions, even 1 year prior to their detection by radiologists,” reported Benoît Audelan, French National Institute for Research in Digital Science and Technology, Université Côte D’Azur, Nice, France.

While CT can detect nodules, it takes multiple scans and technical expertise to determine if the nodule is suspicious. The current algorithm might reduce the time needed to make a definitive diagnosis, said Audelan.

The researchers trained their artificial intelligence program using a set of CT scans from 888 patients that had already been examined by radiologists to identify suspicious growths. They then tested it on a different set of 1,179 patients who were part of a lung screening trial with 3-year follow-up, only using CT scans that were taken in the last 2 years of the trial. These included 177 patients who were diagnosed with lung cancer via a biopsy after their final scan in the trial.

Audelon said the new program identified 172 of the 177 malignant tumours in those CT scans, meaning it was 97% effective in detecting cancers. The 5 tumours that the program missed were near the center of the chest, where tumours are harder to distinguish from healthy parts of the body, he noted.

“Screening for lung cancer would mean many more CT scans being taken and we do not have enough radiologists to review them all,” said Audelon. “That’s why we need to develop computer programs that can help. Our study shows that this program can find possible signs of lung cancer up to a year earlier.”

“The objective of our research is not to replace radiologists, but to assist them by giving them a tool that can spot the earliest signs of lung cancer,” he added.

He said the research team is now in the process of trying to further refine the program to reduce numbers of false positives.

[Presentation title: Validation of Lung Nodule Detection a Year Before Diagnosis in NLST Dataset Based on a Deep Learning System. Abstract OA4317]
Log in to post comments

Tell us what you think of DG News

Click to like Click to dislike