Nearly 40% of breast MRIs without lesions were dismissed by a deep-learning model while no cancerous lesions were missed in women with dense breasts, found a study published October 5, 2021, in Radiology.
“Automated analysis of breast MRI examinations in women with dense breasts dismissed nearly 40% of MRI scans without lesions while not missing any cancers,” wrote lead author Erik Verbung of Utrecht University in the Netherlands. His team of researchers “said the model successfully differentiated between women with and without breast lesions in a study of nearly 4,600 individuals. The authors added that to their knowledge, this study is the first to investigate image-based triaging with use of multicenter screening data of women with extremely dense breasts at average risk.” according to Allegretto.
The team concluded “a deep learning model was developed that identified breast MRI examinations without cancer with high certainty. Using internal-external validation, the method identified nearly 40% of the MRI scans with normal anatomic and physiologic variation in women with extremely dense breasts without missing any malignant disease.”
Erik Verburg, Carla H. van Gils, Bas H. M. van der Velden, Marije F. Bakker, Ruud M. Pijnappel, Wouter B. Veldhuis, and Kenneth G. A. Gilhuijs. Deep Learning for Automated Triaging of 4581 Breast MRI Examinations from the DENSE Trial. Radiology. Published online October 5, 2021, at https://pubs.rsna.org/doi/10.1148/radiol.2021203960. Accessed October 11, 2021.