Skip to content

Robust Breast Cancer Detection in Mammography and Digital Breast Tomosynthesis Using an Annotation-Efficient Deep Learning Approach

January 1, 2021
2 min

Nature Medicine
William Lotter, Abdul Rahman Diab, Bryan Haslam, Jiye G. Kim, Giorgia Grisot, Eric Wu, Kevin Wu, Jorge Onieva Onieva, Yun Boyer, Jerrold L. Boxerman, Meiyun Wang, Mack Bandler, Gopal R. Vijayaraghavan & A. Gregory Sorensen

Abstract 

Breast cancer remains a global challenge, causing over 600,000 deaths in 2018. To achieve earlier cancer detection, health organizations worldwide recommend screening mammography, which is estimated to decrease breast cancer mortality by 20–40%. Despite the clear value of screening mammography, significant false positive and false negative rates along with non-uniformities in expert reader availability leave opportunities for improving quality and access. To address these limitations, there has been much recent interest in applying deep learning to mammography, and these efforts have highlighted two key difficulties: obtaining large amounts of annotated training data and ensuring generalization across populations, acquisition equipment and modalities. Here we present an annotation-efficient deep learning approach that (1) achieves state-of-the-art performance in mammogram classification, (2) successfully extends to digital breast tomosynthesis (DBT; ‘3D mammography’), (3) detects cancers in clinically negative prior mammograms of patients with cancer, (4) generalizes well to a population with low screening rates and (5) outperforms five out of five full-time breast-imaging specialists with an average increase in sensitivity of 14%. By creating new ‘maximum suspicion projection’ (MSP) images from DBT data, our progressively trained, multiple-instance learning approach effectively trains on DBT exams using only breast-level labels while maintaining localization-based interpretability. Altogether, our results demonstrate promise towards software that can improve the accuracy of and access to screening mammography worldwide. 

Disclaimers: 

Breast Suite comprises multiple applications including ProFound Pro, Breast Density, Safeguard Review, Risk Assessment, BAC and DeepHealth Viewer. DeepHealth Viewer is manufactured by eRAD, Inc. and distributed by DeepHealth, Inc.  

Any claims made about Breast Suite may reference claims associated with its individual components. 

Not all products and functionalities are commercially available in all countries. For clearance and commercial availability in your geography of functionalities listed and compatibility with other systems, please contact your account manager. 

Read full article

Disclaimers: DeepHealth Prostate is manufactured as Quantib Prostate by Quantib BV for DeepHealth Inc., DeepHealth Lung is manufactured as Veye Lung Nodules by Aidence BV for DeepHealth Inc., DeepHeath Lung Tracker is manufactured as Veye Clinic by Aidence BV for DeepHealth Inc. and DeepHealth Brain is manufactured as Quantib ND by Quantib BV for DeepHealth Inc. Not all products and functionalities are commercially available in all countries. For clearance and commercial availability in your geography of functionalities listed and compatibility with other systems, please contact us