Can we use an artificial intelligence system to improve the quality and efficiency of breast cancer screening?

Not Applicable

Conditions: Decision support in breast cancer screening
Not Applicable

Registration Number: ISRCTN88754382

Lead Sponsor: Imperial College London

Brief Summary: Not available

Detailed Description: Not available

Recruitment & Eligibility

Status: Ongoing

Sex: Female

Target Recruitment: 10875

Inclusion Criteria

1. Women undergoing routine breast cancer screening (age 50–70), as part of the national breast screening programme at Imperial College Healthcare NHS Trust and St George’s University Hospital NHS Foundation Trust between the study dates.
2. Mammography images acquired using Hologic/Lorad, Siemens, or GE devices.

Exclusion Criteria

1. Women that opt-out of this study
2. Women who have registered with the NHS national data opt-out

Study & Design

Study Type: Observational

Study Design: Not specified

Primary Outcome Measures

Name	Time	Method
1. Time taken for the AI system to return results from mammograph images over the study dataset time period <br>2. Analysis of number of failure cases ( such as such as model errors, software errors, integration errors, use errors, and hardware errors) for the study dataset time period. Accuracy will be measured as proportion of true results (both true positives and true negatives) among whole instances. Area under the receiver operating characteristic curve (ROC) will be measured for AI<br>3. Percentage of cases correctly excluded during eligibility checks and reasons do excursion during the study period

Secondary Outcome Measures

Name	Time	Method
1. Accuracy measures including AI recall rate measured as proportion of true results (both true positives and true negatives) <br>2. AI sensitivity and specificity with respect to arbitrated recall decisions (measured as the number of positive cases (cases considered positive if they received a biopsy-confirmed diagnosis of cancer within 3 months following the screening visit. Negative cases will require a negative result from the study screening visit) <br>3. AI sensitivity for biopsy-proven cancer u(true positive rate in percentage(%) derived by ROC analysis)<br>4. AI specificity for biopsy or diagnostic imaging-proven benign lesions (true negative rate in percentage (%) derived by ROC analysis)