AI-supported approaches for mammography single and double reading: A controlled multireader study.
Authors
Affiliations (32)
Affiliations (32)
- Unit of Breast Imaging, Istituto per lo Studio, la Prevenzione e la Rete Oncologica (ISPRO), Florence, Italy. Electronic address: [email protected].
- Postgraduate School in Radiodiagnostics, Università degli Studi di Milano, Milan, Italy. Electronic address: [email protected].
- Cancer Risk Factors and Lifestyle Epidemiology Unit, Istituto per lo Studio, la Prevenzione e la Rete Oncologica (ISPRO), Florence, Italy. Electronic address: [email protected].
- Unit of Breast Imaging, Istituto per lo Studio, la Prevenzione e la Rete Oncologica (ISPRO), Florence, Italy. Electronic address: [email protected].
- Unit of Breast Imaging, Istituto per lo Studio, la Prevenzione e la Rete Oncologica (ISPRO), Florence, Italy. Electronic address: [email protected].
- Unit of Breast Imaging, Istituto per lo Studio, la Prevenzione e la Rete Oncologica (ISPRO), Florence, Italy. Electronic address: [email protected].
- Unit of Breast Imaging, Istituto per lo Studio, la Prevenzione e la Rete Oncologica (ISPRO), Florence, Italy. Electronic address: [email protected].
- Unit of Breast Imaging, Istituto per lo Studio, la Prevenzione e la Rete Oncologica (ISPRO), Florence, Italy. Electronic address: [email protected].
- Unit of Breast Imaging, Istituto per lo Studio, la Prevenzione e la Rete Oncologica (ISPRO), Florence, Italy. Electronic address: [email protected].
- Unit of Breast Imaging, Istituto per lo Studio, la Prevenzione e la Rete Oncologica (ISPRO), Florence, Italy. Electronic address: [email protected].
- Azienda USL Toscana Centro, Florence, Italy. Electronic address: [email protected].
- Azienda USL Toscana Centro, Florence, Italy. Electronic address: [email protected].
- Azienda USL Toscana Centro, Florence, Italy. Electronic address: [email protected].
- Azienda USL Toscana Centro, Florence, Italy. Electronic address: [email protected].
- Azienda USL Toscana Centro, Florence, Italy. Electronic address: [email protected].
- Azienda USL Toscana Centro, Florence, Italy. Electronic address: [email protected].
- Azienda USL Toscana Centro, Florence, Italy. Electronic address: [email protected].
- Azienda USL Toscana Centro, Florence, Italy. Electronic address: [email protected].
- Azienda USL Toscana Nord-Ovest, Pisa, Italy. Electronic address: [email protected].
- Azienda USL Toscana Nord-Ovest, Pisa, Italy. Electronic address: [email protected].
- Azienda USL Toscana Nord-Ovest, Pisa, Italy. Electronic address: [email protected].
- Azienda USL Toscana Nord-Ovest, Pisa, Italy. Electronic address: [email protected].
- Azienda USL Toscana Nord-Ovest, Pisa, Italy. Electronic address: [email protected].
- Azienda USL Toscana Nord-Ovest, Pisa, Italy. Electronic address: [email protected].
- Azienda Ospedaliero-Universitaria Pisana, Pisa, Italy. Electronic address: [email protected].
- Azienda Ospedaliero-Universitaria Senese, Siena, Italy. Electronic address: [email protected].
- Postgraduate School in Radiodiagnostics, Università degli Studi di Firenze, Florence, Italy. Electronic address: [email protected].
- Postgraduate School in Radiodiagnostics, Università degli Studi di Pisa, Pisa, Italy. Electronic address: [email protected].
- Postgraduate School in Radiodiagnostics, Università degli Studi di Pisa, Pisa, Italy. Electronic address: [email protected].
- Postgraduate School in Radiodiagnostics, Università degli Studi di Pisa, Pisa, Italy. Electronic address: [email protected].
- Imaging Institute of Southern Switzerland (IIMSI), Ente Ospedaliero Cantonale (EOC), Lugano, Switzerland; Faculty of Biomedical Sciences, Università della Svizzera Italiana, Lugano, Switzerland. Electronic address: [email protected].
- Imaging Institute of Southern Switzerland (IIMSI), Ente Ospedaliero Cantonale (EOC), Lugano, Switzerland. Electronic address: [email protected].
Abstract
To assess the impact of artificial intelligence (AI) on the diagnostic performance of radiologists with varying experience levels in mammography reading, considering single and simulated double reading approaches. In this retrospective study, 150 mammography examinations (30 with pathology-confirmed malignancies, 120 without malignancies [confirmed by 2-year follow-up]) were reviewed according to five approaches: A) human single reading by 26 radiologists of varying experience; B) AI single reading (Lunit INSIGHT MMG; C) human single reading with simultaneous AI support; D) simulated human-human double reading; E) simulated human-AI double reading, with AI as second independent reader flagging cases with a cancer probability ≥10 %. Sensitivity and specificity were calculated and compared using McNemar's test, univariate and multivariable logistic regression. Compared to single reading without AI support, single reading with simultaneous AI support improved mean sensitivity from 69.2 % (standard deviation [SD] 15.6) to 84.5 % (SD 8.1, p < 0.001), providing comparable mean specificity (91.8 % versus 90.8 %, p = 0.06). The sensitivity increase provided by the AI-supported single reading was largest in the group of radiologists with a sensitivity below the median in the non-supported single reading, from 56.7 % (SD 12.1) to 79.7 % (SD 10.2, p < 0.001). In the simulated human-AI double reading approach, sensitivity further increased to 91.8 % (SD 3.4), surpassing that of the human-human simulated double reading (87.4 %, SD 8.8, p = 0.016), with comparable mean specificity (from 84.0 % to 83.0 %, p = 0.17). AI support significantly enhanced sensitivity across all reading approaches, particularly benefiting worse performing radiologists. In the simulated double reading approaches, AI incorporation as independent second reader significantly increased sensitivity without compromising specificity.