ProCAN: Progressive growing channel attentive non-local network for lung nodule classification

作者：

Highlights：

• Generalize the non-local network by adding channel attention and spatial attention.

• Enhance the progressive growing algorithm by introducing a Bernoulli matrix.

• Train the model using curriculum learning with the nodule diameter criterion.

• Achieve state-of-the-art results on the public LIDC-IDRI and LUNGx datasets.

摘要

•Generalize the non-local network by adding channel attention and spatial attention.•Enhance the progressive growing algorithm by introducing a Bernoulli matrix.•Train the model using curriculum learning with the nodule diameter criterion.•Achieve state-of-the-art results on the public LIDC-IDRI and LUNGx datasets.

论文关键词：Self-Attention,Non-local network,Nodule classification,Curriculum learning,Deep learning

论文评审过程：Received 14 January 2021, Revised 1 July 2021, Accepted 6 September 2021, Available online 17 September 2021, Version of Record 23 September 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.108309