Stand-Alone Use of Artificial Intelligence for Digital Mammography and Digital Breast Tomosynthesis Screening: A Retrospective Evaluation

Link to the original article

Background

Use Transpara as a stand-alone reader for digital mammography (DM) or digital breast tomosynthesis (DBT) breast screening could ease radiologists’ workload while maintaining quality.

Objectives

Retrospectively evaluating Transpara’s stand-alone performance as an independent reader of DM and DBT screening examinations.

Methods

Consecutive screening-paired and independently read DM and DBT images acquired between January 2015 and December 2016 were retrospectively collected from the Tomosynthesis Cordoba Screening Trial. Transpara computed a cancer risk score (range, 1–100) for DM and DBT examinations independently. Transpara’s stand-alone performance was measured using the area under the receiver operating characteristic curve (AUC) and sensitivity and recall rate at different operating points selected to have noninferior sensitivity compared with the human readings (non-inferiority margin, 5%). The recall rate of Transpara and the human readings were compared using a McNemar test.

Results

A total of 15 999 DM and DBT examinations (113 breast cancers, including 98 screen-detected and 15 interval cancers) from 15 998 women (mean age: 58 years) were evaluated. Transpara achieved an AUC of 0.93 for DM and 0.94 for DBT. For DM, Transpara achieved noninferior sensitivity as a single (58.4%; 66 of 113) or double (67.3%; 76 of 113) reader, with a reduction (p < .001) in recall rate of up to 2%. For DBT, Transpara achieved noninferior sensitivity as a single (77%; 87 of 113) or double (81.4%; 92 of 113) reader, but with a higher recall rate (p < .001) of up to 12.3%.

Conclusion

Transpara could replace radiologists’ readings in breast screening, achieving a noninferior sensitivity, with a lower recall rate for digital mammography but a higher recall rate for digital breast tomosynthesis.