Doubly supervised parameter transfer classifier for diagnosis of breast cancer with imbalanced ultrasound imaging modalities

作者：

Highlights：

• We propose a novel doubly supervised parameter transfer classifier (DSPTC) for transfer learning.

• DSPTC simultaneously perform knowledge transfer between both the paired data with shared labels and the unpaired data with different labels.

• DSPTC effectively improves the performance of single-modal B-mode ultrasound based CAD for breast cancers.

摘要

•We propose a novel doubly supervised parameter transfer classifier (DSPTC) for transfer learning.•DSPTC simultaneously perform knowledge transfer between both the paired data with shared labels and the unpaired data with different labels.•DSPTC effectively improves the performance of single-modal B-mode ultrasound based CAD for breast cancers.

论文关键词：Doubly supervised parameter transfer classifier,Support vector machine plus,Hilbert-Schmidt independence criterion,B-mode ultrasound,Breast cancer

论文评审过程：Received 30 September 2020, Revised 15 March 2021, Accepted 28 June 2021, Available online 18 July 2021, Version of Record 25 July 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.108139