Clinical Scorecard: Assessment of Interval Breast Cancer Screening Mammograms Utilizing Radiologists and Artificial Intelligence: A Retrospective Study

At a Glance

Key Highlights

• Interval breast cancers (IBC) occur within screening intervals and are associated with more aggressive tumors and higher mortality.
• AI systems like ProFound AI® can analyze mammograms to identify signs of breast cancer potentially missed by radiologists.
• Reducing IBC incidence is a key quality indicator for mammography screening programs and may be improved by AI without increasing workload.

Guideline-Based Recommendations

Diagnosis

• Classify breast cancer as interval cancer if diagnosed within 24 months after a normal screening mammogram.
• Use double reading by experienced radiologists to evaluate screening mammograms.
• Consider AI systems to retrospectively identify signs of breast cancer in interval cancer mammograms.

Management

• Invite women aged 50-69 to biennial mammography screening covered by compulsory health insurance with minimal co-payment.
• Exclude benign conditions such as LCIS from invasive breast cancer analyses and management.
• Consider shortening screening intervals or additional imaging modalities (e.g., MRI) for women at higher risk, balancing cost and burden.

Monitoring & Follow-up

• Use the rate of interval breast cancers as a key quality indicator of mammography screening program effectiveness.
• Monitor AI system performance using case scores and risk scores to assess breast cancer likelihood and forecast risk.
• Maintain cancer registries to track diagnosis, staging, treatment, and outcomes for continuous quality assessment.

Risks

• Interval breast cancers may be missed due to dense breast tissue, slow tumor development, or radiologist perceptual/interpretation errors.
• Increasing screening frequency or imaging modalities may increase workload, costs, and patient burden.
• AI integration into routine screening workflows requires consistent evidence of performance, especially in detecting interval cancers.

Patient & Prescribing Data

Women aged 50-69 participating in Swiss mammography screening programs with identified interval breast cancers

Interval breast cancers are associated with more aggressive tumor characteristics and higher rates of invasive treatments compared to screen-detected cancers.

Clinical Best Practices

• Implement double reading of mammograms by experienced radiologists to improve detection accuracy.
• Utilize AI systems like ProFound AI® to support radiologists in identifying subtle signs of breast cancer in screening mammograms.
• Exclude benign lesions such as LCIS from breast cancer diagnosis and management protocols.
• Maintain comprehensive cancer registries for accurate tracking of screening outcomes and interval cancer incidence.
• Consider risk stratification using AI-derived risk scores to tailor screening intervals and modalities.

References

• European guidelines on breast cancer screening and diagnosis
• ProFound AI® system information by iCAD
• STARD guideline for reporting diagnostic accuracy studies