AI proven to safely improve breast cancer detection accuracy, study finds
Artificial intelligence (AI) is a safe and efficient tool for breast cancer detection and boosting the accuracy of diagnoses, a new early-stage study in The Lancet Oncology has found.
The AI in the study was able to cut doctor mammogram screen-reading workload almost in half, by 44.3 per cent.
The study, published Tuesday, does not argue in favour of AI as a replacement for radiologists, but rather for its ability to improve screen-reading speed and accuracy in a safe manner.
While other studies have shown promising results for the use of AI to improve mammography screening accuracy, this study is thought to be the first randomized trial, which randomly assigns participants to an experimental group or control group.
Researchers scanned over 80,000 women aged 40 to 80 years old at four screening sites in Sweden between April 2021 and July 2022.
The women were assigned to either an intervention group, whose mammograms were read by radiologists with the support of AI, or a control group, where mammograms were read by two radiologists without the use of AI.
Cancer was detected at a rate of 6.1 per 1,000 screened participants for the intervention group whose scan readings were supported by AI — or 244 out of a total of 39,996 screen readings.
Detection rates in the control group were 5.1 per 1,000 or 203 out of a total of 40,024 non-AI supported readings.
“AI-supported mammography screening resulted in a similar cancer detection rate compared with standard double reading, with a substantially lower screen-reading workload, indicating that the use of AI in mammography screening is safe,” the study says.
The false positive rate came out at 1.5 per cent for both groups, showing that the additional cancer detections made by AI are not due to an over-sensitivity.
The use of AI in health care and research has been picking up steam, with some promising results.
A study published in January 2020 by the journal Nature also found that AI can more accurately predict breast cancer than humans.
The AI in the study reduced false positives by 5.7 per cent in the U.S. and 1.2 per cent in the U.K. datasets. It also reduced false negatives by 9.4 per cent in the U.S. and 2.7 per cent in the U.K., meaning it picked up on cancers that humans had missed.
Breast cancer is the second leading cause of death from cancer in Canadian women, according to the Canadian Cancer Society. It is estimated that one in eight Canadian women will develop breast cancer during their lifetime and one in 34 will die from it.
A lab out of Waterloo, Ont., is working to help patients get proper treatment with new AI-driven technology.
When patients get breast cancer, they typically undergo imaging such as magnetic resonance imaging or MRI to look for cancerous tumors. The Waterloo lab has created “a synthetic correlate diffusion” MRI that is tailored to capture details and properties of cancer in a way that previous MRI systems couldn’t.
“It could be a very helpful tool to help oncologists and medical doctors to be able to identify and personalize the type of treatment that a cancer patient gets,” Alexander Wong, professor and Canada Research Chair in Artificial Intelligence and Medical Imaging at the University of Waterloo, told Global News in an interview in February.
Using synthetic correlated diffusion imaging data, the new technology predicts whether a patient is likely to benefit from neoadjuvant chemotherapy – or chemotherapy that occurs before surgery.
Though the hardware of the actual MRI machine hasn’t changed in this model, what has altered is the way the technology sends “pulses” through the patient’s body and how it collects data.
“It’s essentially the combination of two types of technologies. One is the new MRI imaging technology to really capture the right information. The other is the AI advancement in terms of a deep neural network,” Wong said.
— With files from Global News’ Irelyne Lavery and Katherine Ward.
by Global News
