Breast imaging readers' performance in the PERFORMS test-set based assessment scheme within the MyPeBS international...

Area of Science:

• Radiology
• Medical Imaging
• Breast Cancer Screening

Background:

• Quality Assurance (QA) practices in breast imaging show significant variation across Europe.
• The UK mandates regular performance monitoring for screen readers.
• Assessing diagnostic performance variability among readers is crucial for effective screening.

Purpose of the Study:

• To evaluate the variability in diagnostic performance among breast imaging readers in a multinational prospective trial.
• To identify potential outliers in cancer detection and return-to-screen rates.

Main Methods:

• A retrospective multinational study involving 110 breast imaging readers from 6 countries.
• Readers analyzed a test set of 40 challenging breast screening cases using PERFORMS software.
• Cancer detection and correct return-to-screen rates were calculated for each participant.

Main Results:

• Significant variability observed in cancer detection rates (mean 73.6%, SD 19.7%) and correct return-to-screen rates (mean 79.7%, SD 10.5%).
• Outlier readers with extremely low performance were identified (2.7% for cancer detection, 1.8% for return-to-screen).
• Data collected from 88 readers via an accompanying questionnaire on screening work and experience.

Conclusions:

• Test set assessments like PERFORMS can reflect real-world breast screening proficiency.
• The identification of outlier readers underscores the necessity for double reading.
• Standardized QA protocols are essential for patient safety and service efficiency in breast cancer screening.