Abstract
Purpose: To compare the performance of a Digital Breast Tomosynthesis (DBT) Artificial Intelligence (AI) model as a standalone reader to that of a large cohort of breast imaging readers, using the Personal Performance in Mammographic Screening (PERFORMS) scheme.
Methods: 75 challenging combined DBT and Synthetic 2D mammography (S2D) screening cases were collated into a PERFORMS test-set. Test-set images were analysed by a prototype server allowing batch-processing of a commercial AI model (Hologic Genius AI® Detection v2.0). The set was also distributed to 96 readers from 7 UK National Health Service (NHS) hospitals that use DBT in screening as part of the PROSPECTS trial, and to 6 readers from 1 US institution that employs DBT in routine screening. The AI performance will be benchmarked against the performance of this reader cohort.
Results: The AI model achieved an Area Under the Receiver Operating Characteristic Curve (AUC) of 0.935, and a sensitivity of 89.5% and specificity of 85.7% at the optimal threshold (=33). At the time of submission, 35 human readers from the cohort have completed the test-set, with a mean human sensitivity of 90.8% (SD: 8.8%) and specificity of 85.2% (SD: 16.1%). Currently, there are another 42 readers completing the test-set; their data will be collated with the current cohort and the complete cohort will be reported at the conference.
Conclusion: This international, Multiple Reader Multiple Case (MRMC) study enables the comparison of a very large cohort of breast imaging readers to a DBT AI model.
[Poster P41. Awarded winner of best poster presentation.]
Methods: 75 challenging combined DBT and Synthetic 2D mammography (S2D) screening cases were collated into a PERFORMS test-set. Test-set images were analysed by a prototype server allowing batch-processing of a commercial AI model (Hologic Genius AI® Detection v2.0). The set was also distributed to 96 readers from 7 UK National Health Service (NHS) hospitals that use DBT in screening as part of the PROSPECTS trial, and to 6 readers from 1 US institution that employs DBT in routine screening. The AI performance will be benchmarked against the performance of this reader cohort.
Results: The AI model achieved an Area Under the Receiver Operating Characteristic Curve (AUC) of 0.935, and a sensitivity of 89.5% and specificity of 85.7% at the optimal threshold (=33). At the time of submission, 35 human readers from the cohort have completed the test-set, with a mean human sensitivity of 90.8% (SD: 8.8%) and specificity of 85.2% (SD: 16.1%). Currently, there are another 42 readers completing the test-set; their data will be collated with the current cohort and the complete cohort will be reported at the conference.
Conclusion: This international, Multiple Reader Multiple Case (MRMC) study enables the comparison of a very large cohort of breast imaging readers to a DBT AI model.
[Poster P41. Awarded winner of best poster presentation.]
| Original language | English |
|---|---|
| Publication status | Published - 9 Jul 2025 |
| Event | Symposium Mammographicum 2025 - Bournemouth, United Kingdom Duration: 9 Jul 2025 → 11 Jul 2025 https://sympmamm.org.uk/ |
Conference
| Conference | Symposium Mammographicum 2025 |
|---|---|
| Country/Territory | United Kingdom |
| City | Bournemouth |
| Period | 9/07/25 → 11/07/25 |
| Internet address |
UN SDGs
This output contributes to the following UN Sustainable Development Goals (SDGs)
-
SDG 3 Good Health and Well-being
Fingerprint
Dive into the research topics of 'Evaluation of a digital breast tomosynthesis cancer detection ai algorithm using the Personal Performance in Mammographic Screening Scheme (PERFORMS)'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver