Breast cancer is a prevalent disease affecting approximately one in seven Australian women during their lifetime, resulting in around 20,000 new diagnoses each week. Mammograms have long been an essential tool for detecting early-stage breast cancer by utilizing X-ray imaging techniques. However, a recent Swedish study published in The Lancet has shown that incorporating artificial intelligence (AI) into mammogram analysis can enhance cancer detection rates by 20% and decrease the workload by 44%.

Currently, mammograms are assessed by two radiologists who determine if any abnormalities are present. If an anomaly is detected, further diagnostic tests, such as additional mammograms, ultrasounds, needle biopsies, or surgery, may be recommended at a BreastScreen assessment clinic. While most women referred for additional tests are ultimately cleared of cancer, approximately one in ten receive a breast cancer diagnosis.

The introduction of AI-supported mammogram reading systems has the potential to improve efficiency by directing radiologists’ attention towards suspicious areas on the images. The Swedish study, which involved 80,000 women aged 40-80 participating in a screening program, found that AI-assisted mammogram analysis resulted in a substantial reduction in radiologists’ workload. When AI did not identify any suspicious areas, only one radiologist had to review the mammogram, saving valuable time.

Additionally, the use of AI appeared to enhance reading accuracy. The AI-assisted group had a slightly higher referral rate for further assessment compared to the standard double reading group, leading to the detection of more cancers. Specifically, 28% of women in the AI-supported group were diagnosed with cancer, compared to 25% in the standard reading group. This equated to an increase of one additional cancer detected for every 1,000 women screened.

Despite these promising results, it is essential to address the risk of overdiagnosis. While AI detection may identify more cancers, there is a concern that some of these smaller tumors may not pose a significant threat. Additionally, it remains unclear if AI can effectively detect faster-growing aggressive cancers that arise between mammograms, known as “interval cancers.”

Further research and data analysis are necessary before implementing AI into screening programs. Ongoing studies, including data collected in Australia, will provide more mature data and insights. Ultimately, the goal is to improve cancer survival rates and alleviate the burden of treatment for affected individuals.

The post Artificial Intelligence in Breast Cancer Detection: A Promising Advancement appeared first on TS2 SPACE.