PURPOSE: To develop a Breast Imaging Reporting and Data System (BI-RADS) breast density deep learning (DL) model in a multisite setting for synthetic two-dimensional mammographic (SM) images derived from digital breast tomosynthesis examinations by using full-field digital mammographic (FFDM) images and limited SM data. MATERIALS AND METHODS: A DL model was trained to predict BI-RADS breast density by using FFDM images acquired from 2008 to 2017 (site 1: 57 492 patients, 187 627 examinations, 750 752 images) for this retrospective study. The FFDM model was evaluated by using SM datasets from two institutions (site 1: 3842 patients, 3866 examinations, 14 472 images, acquired from 2016 to 2017; site 2: 7557 patients, 16 283 examinations, 63 973 images, 2015 to 2019). Each of the three datasets were then split into training, validation, and test. Adaptation methods were investigated to improve performance on the SM datasets, and the effect of dataset size on each adaptation method was considered. Statistical significance was assessed by using CIs, which were estimated by bootstrapping. RESULTS: Without adaptation, the model demonstrated substantial agreement with the original reporting radiologists for all three datasets (site 1 FFDM: linearly weighted Cohen κ [κw] = 0.75 [95% CI: 0.74, 0.76]; site 1 SM: κw = 0.71 [95% CI: 0.64, 0.78]; site 2 SM: κw = 0.72 [95% CI: 0.70, 0.75]). With adaptation, performance improved for site 2 (site 1: κw = 0.72 [95% CI: 0.66, 0.79], 0.71 vs 0.72, P = .80; site 2: κw = 0.79 [95% CI: 0.76, 0.81], 0.72 vs 0.79, P < .001) by using only 500 SM images from that site. CONCLUSION: A BI-RADS breast density DL model demonstrated strong performance on FFDM and SM images from two institutions without training on SM images and improved by using few SM images.Supplemental material is available for this article.Published under a CC BY 4.0 license. 2020 by the Radiological Society of North America, Inc.
PURPOSE: To develop a Breast Imaging Reporting and Data System (BI-RADS) breast density deep learning (DL) model in a multisite setting for synthetic two-dimensional mammographic (SM) images derived from digital breast tomosynthesis examinations by using full-field digital mammographic (FFDM) images and limited SM data. MATERIALS AND METHODS: A DL model was trained to predict BI-RADS breast density by using FFDM images acquired from 2008 to 2017 (site 1: 57 492 patients, 187 627 examinations, 750 752 images) for this retrospective study. The FFDM model was evaluated by using SM datasets from two institutions (site 1: 3842 patients, 3866 examinations, 14 472 images, acquired from 2016 to 2017; site 2: 7557 patients, 16 283 examinations, 63 973 images, 2015 to 2019). Each of the three datasets were then split into training, validation, and test. Adaptation methods were investigated to improve performance on the SM datasets, and the effect of dataset size on each adaptation method was considered. Statistical significance was assessed by using CIs, which were estimated by bootstrapping. RESULTS: Without adaptation, the model demonstrated substantial agreement with the original reporting radiologists for all three datasets (site 1 FFDM: linearly weighted Cohen κ [κw] = 0.75 [95% CI: 0.74, 0.76]; site 1 SM: κw = 0.71 [95% CI: 0.64, 0.78]; site 2 SM: κw = 0.72 [95% CI: 0.70, 0.75]). With adaptation, performance improved for site 2 (site 1: κw = 0.72 [95% CI: 0.66, 0.79], 0.71 vs 0.72, P = .80; site 2: κw = 0.79 [95% CI: 0.76, 0.81], 0.72 vs 0.79, P < .001) by using only 500 SM images from that site. CONCLUSION: A BI-RADS breast density DL model demonstrated strong performance on FFDM and SM images from two institutions without training on SM images and improved by using few SM images.Supplemental material is available for this article.Published under a CC BY 4.0 license. 2020 by the Radiological Society of North America, Inc.
Authors: Said Pertuz; Elizabeth S McDonald; Susan P Weinstein; Emily F Conant; Despina Kontos Journal: Radiology Date: 2015-10-21 Impact factor: 11.105
Authors: Elizabeth A Rafferty; Jeong Mi Park; Liane E Philpotts; Steven P Poplack; Jules H Sumkin; Elkan F Halpern; Loren T Niklason Journal: AJR Am J Roentgenol Date: 2014-02 Impact factor: 3.959
Authors: Kathleen R Brandt; Christopher G Scott; Lin Ma; Amir P Mahmoudzadeh; Matthew R Jensen; Dana H Whaley; Fang Fang Wu; Serghei Malkov; Carrie B Hruska; Aaron D Norman; John Heine; John Shepherd; V Shane Pankratz; Karla Kerlikowske; Celine M Vachon Journal: Radiology Date: 2015-12-22 Impact factor: 11.105
Authors: Brian L Sprague; Emily F Conant; Tracy Onega; Michael P Garcia; Elisabeth F Beaber; Sally D Herschorn; Constance D Lehman; Anna N A Tosteson; Ronilda Lacson; Mitchell D Schnall; Despina Kontos; Jennifer S Haas; Donald L Weaver; William E Barlow Journal: Ann Intern Med Date: 2016-07-19 Impact factor: 25.391
Authors: Paul H Yi; Abigail Lin; Jinchi Wei; Alice C Yu; Haris I Sair; Ferdinand K Hui; Gregory D Hager; Susan C Harvey Journal: J Digit Imaging Date: 2019-08 Impact factor: 4.056
Authors: Aimilia Gastounioti; Shyam Desai; Vinayak S Ahluwalia; Emily F Conant; Despina Kontos Journal: Breast Cancer Res Date: 2022-02-20 Impact factor: 8.408