In a groundbreaking clinical trial, researchers are exploring the implementation of artificial intelligence (AI) in mammography and digital breast tomosynthesis (DBT) for breast cancer screening. The study, approved by the Institutional Review Board at Reina Sofía University Hospital in Córdoba, Spain, aims to determine whether AI can enhance the accuracy and efficiency of breast cancer detection although reducing the workload for radiologists.
Conducted within the framework of the Andalusian Breast Cancer Screening Program, the trial invites women aged 50 to 71 to participate in screenings every two years. The aim is to evaluate the effectiveness of AI-assisted readings compared to the traditional double reading by radiologists without AI support. This paired noninferiority trial design will help ascertain if the AI system can deliver comparable results while alleviating some of the pressures on healthcare resources.
From March 2022 to January 2024, all eligible women in the defined age group were invited to enroll, provided they met specific inclusion criteria. Participants were excluded if they showed signs of breast cancer, had breast prostheses, or if their imaging data could not be processed by the AI system due to technical limitations.
Trial Design and Methodology
Each participant will undergo both a standard mammography exam and an AI-assisted reading, employing a commercially available AI system called Transpara (version 1.7) from ScreenPoint Medical. This system is designed to automatically detect and classify regions within breast images that may indicate cancer, generating a suspiciousness score for each region. The AI will assist in evaluating cases where the likelihood of cancer is higher, specifically those scored between 8 and 10 on a scale of 1 to 10.
The study employs four mammography devices, including three digital mammography (DM) units and one DBT unit. Each examination will be read independently by radiologists, ensuring a thorough analysis without consensus or arbitration. The trial’s primary outcomes include cancer detection rates (CDR), recall rates (RR), and overall workload for the radiologists.
Key Findings and Safety Measures
The AI system has undergone extensive evaluation in previous studies, demonstrating performance levels comparable to human radiologists. The findings suggest that AI can improve accuracy in detecting breast cancer and potentially enhance the screening process by reducing the need for double readings in low-risk cases.
To ensure participant safety, all mammographic images will be anonymized before analysis, complying with data protection regulations. The ethical committee has reviewed the study protocol, concluding that it poses minimal risk to participants.
Impact and Future Directions
This trial is poised to significantly influence breast cancer screening practices. By integrating AI into the diagnostic workflow, the study aims to enhance early detection rates while optimizing radiologist efficiency. The results could pave the way for broader implementation of AI technologies in various medical imaging disciplines.
As the trial progresses, stakeholders in the healthcare community will be closely monitoring outcomes related to the CDR and RR of the AI-assisted strategies compared to traditional methods. Should the findings support the hypothesis of noninferiority, it may lead to a paradigm shift in breast cancer screening protocols across various healthcare settings.
For more information about the study and its implications for breast cancer screening, interested parties can access the detailed study protocol and participant dataset on the clinical trial registry and Zenodo platform.
This content is for informational purposes only and should not be considered professional medical advice. Readers are encouraged to consult healthcare professionals for personalized guidance.
