This preliminary study aimed to differentiate malignant from benign enhancing foci on breast MRI using radiomic signature. Forty-five patients were included in the study, with 12 malignant lesions and 33 benign lesions. The study showed how feasible a radiomic approach was in the characterization of enhancing foci on breast MRI.
Key points
- Radiomic signature could distinguish malignant from benign enhancing foci on magnetic resonance imaging of the breast.
- In this study, we applied a “training with input selection and testing “machine learning algorithm on 45 foci, using 8 confirmed benign lesions and 15 confirmed malignant lesions as reference cases.
- Over 200 radiomic features were extracted.
- Overall, a k-nearest neighbour classifier based on 35 selected features showed an over 90% accuracy.
Article: A machine learning approach for differentiating malignant from benign enhancing foci on breast MRI
Authors: Natascha C. D’Amico, Enzo Grossi, Giovanni Valbusa, Francesca Rigiroli, Bernardo Colombo, Massimo Buscema, Deborah Fazzini, Marco Ali, Ala Malasevschi, Gianpaolo Cornalba & Sergio Papa
