Deep learning with multiresolution handcrafted features for brain MRI segmentation

作者：

Highlights：

• NSCT and NSST coefficients can effectively enrich CNN’s features.

• The entropy is an efficient measure to select the useful NSCT and NSST coefficients.

• Directional NSCT and NSST features are beneficial to enhance the bone segmentation.

• NSST directional features provide more details than NSCT directional coefficients.

摘要

•NSCT and NSST coefficients can effectively enrich CNN’s features.•The entropy is an efficient measure to select the useful NSCT and NSST coefficients.•Directional NSCT and NSST features are beneficial to enhance the bone segmentation.•NSST directional features provide more details than NSCT directional coefficients.

论文关键词：CNN,Contourlet,Shearlet,Segmentation,MR

论文评审过程：Received 28 November 2021, Revised 28 June 2022, Accepted 9 July 2022, Available online 14 July 2022, Version of Record 26 July 2022.

论文官网地址：https://doi.org/10.1016/j.artmed.2022.102365