Recent leaps forward in the effectiveness of machine learning technology could change the face of cancer. Two new studies have demonstrated the system’s potential to spot and understand tumors in lung and breast cancer diagnostics just as accurately as experts.
One, published in the Journal of the National Cancer Institute, suggests that artificial intelligence (AI) could reduce breast cancer mortality by allowing more digital mammography scans to be evaluated accurately.
“Before we could decide what is the best way for AI systems to be introduced in the realm of breast cancer screening with mammography, we wanted to know how good these systems can really be,” says Ioannis Sechopoulos, one of the study’s authors.1
The work compared the performance of 101 radiologists to that of a commercially available AI system.
They used 2,652 exams, 653 of which were malignant. The scans were taken from retrospective sources and had been produced by technology from four different mammogram manufacturers. The team found that the AI’s cancer detection accuracy rate was comparable to the average rate of the 101 radiologists.1
Sechopoulos adds: “It was exciting to see that these systems have reached the level of matching the performance of, not just radiologists, but of radiologists who spend at least a substantial portion of their time reading screening mammograms.”1
Uncovering tumor types the AI way
Dartmouth-Hitchcock Medical Center has also published its AI work, which focuses on the technology’s ability to determine tumor patterns and subtypes in lobectomy slides. The team, led by Saeed Hassanpour PhD, developed a machine learning network and found it performed on par with three practicing pathologists.
“Our study demonstrates that machine learning can achieve high performance on a challenging image classification task and has the potential to be an asset to lung cancer management,” notes Hassanpour.2 The team also want to carry out tests in breast, esophageal and colorectal cancer and have made their AI code publicly available in a bid to promote new research and collaboration.
“If validated through clinical trials, our neural network model can potentially be implemented in clinical practice to assist pathologists,” notes Hassanpour, whose work was published in Scientific Reports.
“Our machine learning method is also fast and can process a slide in less than one minute, so it could help triage patients before examination by physicians and potentially greatly assist pathologists in the visual examination of slides.”
- Study finds robots can detect breast cancer as well as radiologists. (2019, March 5). Retrieved from https://www.eurekalert.org/pub_releases/2019-03/oupu-sfr030519.php
- A new machine learning model can classify lung cancer slides at the pathologist level. (2019, March 4). Retrieved from https://www.eurekalert.org/pub_releases/2019-03/dmc-anm030119.php