The Future of Breast Cancer Screening: Current Modalities, AI Integration, and Economic Impacts
Breast cancer screening has undergone significant advancements in recent years, driven by technological innovations and the potential integration of artificial intelligence (AI). These advancements aim to improve early detection rates, reduce false positives, and ultimately save lives. Dr. Debra L. Monticciolo, a former president of the Society of Breast Imaging and the American College of Radiology, highlighted several cutting-edge technologies that are shaping the future of breast imaging. One such technology is contrast-enhanced mammography (CEM), which adds a physiological component to traditional mammography. CEM is gaining attention for its accessibility, speed, and cost-effectiveness compared to magnetic resonance imaging (MRI). However, MRI remains the gold standard for detecting breast cancer due to its high sensitivity, despite its higher costs and the need for contrast injection.
Dr. Monticciolo expressed optimism about ongoing research into non-contrast MRI or alternative contrast agents, which could simplify breast cancer screening and make it more accessible. Another supplemental screening tool that has been explored is breast ultrasound, particularly automated breast ultrasound (ABUS). ABUS offers the advantage of no radiation exposure, but it has not seen widespread adoption due to several limitations. The introduction of digital breast tomosynthesis (DBT) has made ultrasound less favorable as a supplemental screening tool. DBT can detect half of the cancers that ultrasound finds after a negative mammogram, making it a more efficient option. However, ultrasound is still plagued by a high number of false positives, leading to unnecessary biopsies and increased patient anxiety.
Despite the challenges, ABUS was designed to reduce labor and increase reproducibility, but it has not fully addressed the issues of false positives. The rise of DBT has also led to an overwhelming amount of data for radiologists to analyze, creating a need for innovative solutions to manage this information. This is where AI has the potential to make a significant impact. AI can help triage large volumes of images, allowing radiologists to focus on high-risk cases or specific image slices. While AI for breast imaging is still in its early stages, it requires continuous validation on large and diverse datasets to ensure its effectiveness and reliability.
AI can also assist in risk assessment by identifying women who may benefit from supplemental screening, such as MRI. By improving efficiency and accuracy in breast cancer screening, AI can play a crucial role in image analysis, patient triage, and decision-making for supplemental screening. However, human expertise remains essential to ensure accurate diagnoses. The integration of AI into breast imaging could help address the growing shortage of breast imaging specialists and enhance cancer detection rates. The American College of Radiology has made recommendations to the Centers for Medicare & Medicaid Services (CMS) regarding payment rates for these new technologies, highlighting their importance in modern healthcare.
Recent research has demonstrated the potential of AI software to assist in identifying missed interval cancers on screening mammograms. A study led by Muzna Nanaa, PhD, and senior author Prof. Fiona Gilbert of the University of Cambridge found that an AI algorithm could detect nearly one in four missed interval cancers at a high specificity threshold. The software also correctly pinpointed the location of these cancers in almost three out of four cases. However, at lower specificity thresholds, more interval cancers were detected at the expense of increased recall rates. The study aimed to assess how well the AI software localized interval cancers based on cancer category and histopathologic characteristics.
The researchers retrospectively applied a commercial AI algorithm called Insight MMG v 1.1.2.0 to 2,052 screening mammograms interpreted by two human readers. Of these mammograms, 1,548 were normal, and 514 had interval cancers. The software analyzes digital screening mammograms and provides a likelihood score for cancer, along with a heatmap showing the location of suspicious lesions. The authors emphasized the importance of minimizing false-positive heatmaps, as they can increase reading time and distract from recognizing true cancer areas. Previous studies have shown that readers can experience prompt fatigue when faced with excessive false positives.
While the AI software’s cancer localization performance did not vary by tumor histologic type, it had a higher median score for invasive cancers and high-grade cancers. The software correctly identified a lower proportion of true-negative interval cancers compared to interval cancers with minimal signs of malignancy and false-negative interval cancers. It also had a higher proportion of successful localizations for node-positive cancers compared to node-negative cancers. None of the other cancer characteristics studied were associated with a median score above 80. The authors suggest that a threshold can guide clinicians in deciding whether to recall a woman for further screening if her mammogram has a high AI score but no discernible signs of malignancy.
Further studies are needed to examine whether true-negative mammograms, on which interval cancers were not detected by the AI system, are detectable by other screening methods. These findings demonstrate the potential for AI to aid in detecting missed interval cancers on screening mammograms. However, further research is needed to fully understand its impact on cancer diagnosis and management. Another promising development in AI-based breast cancer screening is the use of a tool called Mirai, which interprets data from mammograms to estimate a woman’s short-term risk of cancer incidence. A new study suggests that implementing an AI-based breast cancer screening program could save a health system $109 million annually.
Mirai is open-source and freely available for research, and it is believed to be the best AI model for short-term breast cancer risk assessment. However, the economic value of using this model to guide screening has not been critically evaluated. The study aimed to assess the cost-effectiveness of integrating this risk-stratified screening program into the UK national breast cancer screening program. Researchers developed a model to estimate the impact of different screening intervals on quality of life, cancer survival rates, and costs. The model compared intervals of 1 to 6 years with the current triennial approach.
Implementing a screening schedule of 6 years for low-risk individuals, biannual screening for those at below-average risk, and annual screening for high-risk individuals could save the NHS between $77.3 million to $109.2 million annually. This approach would also increase quality-adjusted life years, with values ranging from $25,600 to $38,400. Even without investing any new resources, this approach could save the NHS $13.6 million annually. The study suggests that risk-stratified screening could reduce the number of screenings, NHS costs, and invasive cancer cases while increasing quality-adjusted life years. This could also free up resources to address screening backlogs and reduce wait times, potentially improving breast cancer outcomes.
The study also outlines potential limitations. Before the final proposal, the American College of Radiology provided recommendations on payment rates for five out of six of the new codes. These new codes aim to acknowledge the additional burden for providers who accept patients requiring these screenings. The new images were captured at the European Synchrotron Radiation Facility and are being called Google Earth for the human heart. The images were obtained using hierarchical phase-contrast tomography. One specialist describes it as an innovative tool for diagnosing heart diseases. This screening program could improve outcomes and potentially save money for the health system.
The market for artificial intelligence in breast imaging is expanding rapidly, with its value estimated to be $423.9 million in 2023 and projected to reach $1886.4 million by 2032. The market is expected to grow at an annual rate of 16.1%. Key players in this market include companies leveraging AI technology to revolutionize breast imaging. AI technology uses machine learning and deep learning algorithms to analyze mammograms with high accuracy. This helps reduce false positives and improves early detection rates for breast cancer, which is a leading cause of death among women globally. One of the biggest advantages of AI in breast imaging is its ability to lessen the workload of radiologists.
AI systems can automatically identify possible abnormalities, allowing radiologists to focus on cases that require their expertise. This improves efficiency and prevents burnout. Due to the increasing prevalence of breast cancer, there is a high demand for advanced AI solutions. AI can provide more precise and timely diagnoses, resulting in better patient outcomes. By analyzing complex imaging data, AI systems can detect subtle signs of cancer that may be missed by traditional methods. As AI technology continues to advance, we can expect to see more innovative applications in breast imaging. The integration of AI into clinical workflows can transform the way breast cancer is screened, diagnosed, and treated.
The increasing number of mammograms and the shortage of breast imaging specialists have created a need for innovative solutions, making AI a promising tool. AI systems can assist radiologists in managing their workload, prioritizing cases, and enhancing diagnostic accuracy. Advanced AI technologies like deep learning and machine learning are driving the development of more sophisticated breast imaging solutions. By incorporating AI, healthcare providers can improve the efficiency of the screening process, reduce false positives, and improve patient outcomes. North America holds the largest market share for AI in breast imaging, thanks to its advanced healthcare infrastructure and investments in research and development. The Asia Pacific region is the second fastest-growing market for AI in breast imaging due to improved healthcare facilities and growing awareness of breast cancer.