Deep learning of longitudinal mammogram examinations for breast cancer risk prediction

作者：

Highlights：

• A novel deep learning model, LRP-NET, is proposed to capture spatiotemporal breast tissue changes over longitudinal negative/benign mammogram examinations for breast cancer risk prediction.

• The structure of LRP-NET is designed referring to clinical knowledge to capture bilateral breast tissue changes in relation to breast cancer risk assessment.

• The LRP-NET model is compared to other related models and shows outperforming performance for near-term breast cancer risk prediction.

摘要

•A novel deep learning model, LRP-NET, is proposed to capture spatiotemporal breast tissue changes over longitudinal negative/benign mammogram examinations for breast cancer risk prediction.•The structure of LRP-NET is designed referring to clinical knowledge to capture bilateral breast tissue changes in relation to breast cancer risk assessment.•The LRP-NET model is compared to other related models and shows outperforming performance for near-term breast cancer risk prediction.

论文关键词：Breast cancer,Risk prediction,Deep learning,Digital mammogram,Longitudinal data

论文评审过程：Received 17 December 2020, Revised 19 March 2022, Accepted 21 July 2022, Available online 22 July 2022, Version of Record 1 August 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108919