DSRD-Net: Dual-stream residual dense network for semantic segmentation of instruments in robot-assisted surgery

作者：

Highlights：

• DSRD-Net can extract multiscale context features of surgical instruments.

• The proposed residual dense combination efficiently resolves the latency problem.

• A minimal number of layers are used to prevent loss with small number of parameters.

• Our trained model and code are publicly available.

摘要

•DSRD-Net can extract multiscale context features of surgical instruments.•The proposed residual dense combination efficiently resolves the latency problem.•A minimal number of layers are used to prevent loss with small number of parameters.•Our trained model and code are publicly available.

论文关键词：Surgical instruments segmentation,Minimally invasive surgery,Gastrointestinal endoscopy and abdominal porcine procedures,DSRD-Net

论文评审过程：Received 30 December 2021, Revised 31 March 2022, Accepted 25 April 2022, Available online 27 April 2022, Version of Record 28 April 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117420