Every year, at least 20,000 Australian women are diagnosed with breast cancer, and tragically, more than 3,300 lose their lives to this disease. Early detection is key in saving women’s lives, making effective breast screening imperative. Studies indicate that participating in regular breast screening can reduce a woman’s risk of dying from breast cancer by half.
In light of this, a recent study published in *The Lancet Digital Health* unveils promising advancements involving artificial intelligence (AI) in breast cancer screening, hinting at a brighter future in early detection methods.
Current Breast Cancer Screening Methods
Since 1992, Australia has provided free mammograms, or breast X-rays, every two years for women aged 50 to 74. However, participation remains just above 50%, which is concerning given the statistics. Alarmingly, about 25% of women diagnosed with breast cancer are found to have the disease between biennial screens. These “interval cancers” are often more aggressive and carry a higher risk of fatality. A more sensitive screening test could potentially catch these cases earlier, improving outcomes.
The Emergence of AI in Screening Procedures
The BreastScreen program was initiated in response to various significant clinical trials conducted between the 1960s and 1980s, yet the screening technology has largely remained unchanged. Recent research is now looking into risk-adjusted screening, which personalizes screening plans based on individual risk factors. This could involve utilizing various technologies for women at a heightened risk of developing breast cancer.
Currently, women’s cancer risks are typically assessed through questionnaires that pinpoint any associated risk factors. One notable factor is breast density, a term that represents the amount of glandular tissue in the breast. High breast density not only increases cancer risk but also complicates detection during a mammogram.
Additionally, one-time genetic testing serves as another tool to identify women at greater lifetime risk for breast cancer. By examining for high-risk gene mutations like BRCA1 and BRCA2, healthcare providers can better estimate an individual’s long-term risk of developing the disease. Recently, there has been a shift toward integrating AI technologies into assessing breast cancer risk, with a recent study highlighting an AI tool known as BRAIx.
What the Study Revealed
This groundbreaking study utilized the AI tool BRAIx, trained on data from BreastScreen Australia, to assist radiologists in evaluating mammograms effectively. It aimed to predict women’s risk of developing breast cancer over a four-year span, focusing on those who had received clear mammogram results.
Among the 95,823 Australian women evaluated, 1.1% (1,098) developed breast cancer within four years following a clear mammogram. In a separate group from Sweden, 6.9% of 4,430 women developed the disease within two years of receiving clear screening results. The findings indicate that BRAIx scores significantly aid in identifying women likely to develop cancer in the one to two-year window following an initially clear screen. Interestingly, the results from the Australian dataset also imply that BRAIx could identify cancers up to three to four years later, albeit with slightly less accuracy.
The implications of these findings are significant; BRAIx could effectively pinpoint women who may benefit from further examinations, such as MRIs or contrast-enhanced mammograms, both of which enhance visibility and detection rates of cancers that traditional mammograms might miss.
Limitations of the Study
As with any study, limitations do exist. Here are two noteworthy considerations:
-
Comparing BRAIx to genetic testing presents challenges. While BRAIx is specifically trained to detect missed or emerging cancers within a four-year time frame, genetic testing estimates a person’s lifetime risk of developing cancer, making direct comparison challenging.
-
Data collection regarding breast density might not meet ideal standards. Although the study demonstrated BRAIx’s superior accuracy over breast density-based assessments, the density data utilized was acquired through a different tool than those employed in the BreastScreen program. Therefore, this finding warrants careful interpretation.
The Path Ahead
The study contributes to a growing body of research indicating that AI risk assessments could revolutionize breast screening programs by facilitating earlier cancer detections. Currently, BRAIx is undergoing trials in the BreastScreen Victoria program to assist with mammogram evaluations, while other Australian states are also adopting various AI tools in their mammogram readings.
Given these promising developments, it may be time for a comprehensive, national evaluation of new AI screening tools. By embracing a more risk-adjusted methodology to breast cancer screening, Australia could potentially enhance early detection and, ultimately, save more lives.
Inspired by: Source
