AI Breast Cancer Screening Detects 20% More Cases, Reduces Wait Times
A groundbreaking study conducted at Lund University confirms the effectiveness of AI-powered breast cancer screening in not only detecting more cases of cancer but also significantly reducing the burden on radiologists. The research, known as the Mammography Screening with Artificial Intelligence (MASAI) trial, demonstrates the promising potential of AI as a viable alternative to conventional screening methods.
Currently, approximately one million women undergo mammography screening annually in Sweden. Each examination requires review by two breast radiologists, which can be time-consuming and contribute to a growing backlog of cases. This delay poses a risk to patients, prompting specialists to explore the benefits of AI breast cancer screening.
The MASAI trial, a pioneering randomized controlled study, enrolled 80,033 women who were randomly divided into two groups. The intervention group consisted of 40,003 participants who underwent AI breast cancer screening, while the control group, comprising 40,030 women, underwent standard double reading without AI assistance.
Led by Kristina Lång, an associate professor in diagnostic radiology at Lund University, the MASAI trial yielded promising results. The application of AI enabled the identification of 20% more cancer cases compared to traditional screening methods, all while maintaining accuracy and avoiding false positives. False positives occur when a woman is recalled for further evaluation but is ultimately cleared of suspicion.
The total number of screen readings reduced significantly with AI-supported screening, alleviating the workload of radiologists by a remarkable 44%. Compared to traditional screening, which involves an extensive volume of 83,231 screen readings, AI-enabled screening involved only 46,345. This improvement in screening efficiency is crucial, as radiologists typically read 50 screening examinations per hour. According to predictions, AI breast cancer screening could potentially reduce screen reading times by up to five months, thereby significantly reducing wait times for millions of breast cancer patients worldwide and improving treatment outcomes.
Moving forward, the research team aims to examine the types of cancers detected with and without the assistance of AI. The primary outcome of the trial will be the evaluation of the interval cancer rate. Interval cancers are those diagnosed between screenings and often have a worse prognosis than cancers detected during screenings.
To assess the interval cancer rate, the trial participants will have a minimum follow-up of two years. This analysis will determine whether AI-supported screening contributes to a more precise and effective breast cancer screening program.
While the adoption of AI-supported screening presents a promising advancement in breast cancer detection, experts emphasize the need to strike a delicate balance between benefits and potential harm. The key objective is to identify clinically significant cancers at early stages, but this must be weighed against the risks associated with false positives and overdiagnosis of slow-growing cancers.
Preliminary analysis indicates that AI-supported screening is safe, as the cancer detection rate remained unaffected despite the significant decrease in radiologists’ screen-reading workload.
The groundbreaking MASAI trial offers hope for a more efficient and accurate breast cancer screening process. As further research and analysis are conducted, AI breast cancer screening could revolutionize the field, bringing immense benefits to both patients and healthcare providers alike.
