qgking / DASC_COVID19

Code for "Domain adaptation based self-correction model for COVID-19 infection segmentation in CT images"

DASC_COVID19

Domain adaptation based self-correction model for COVID-19 infection segmentation in CT images(https://doi.org/10.1016/j.eswa.2021.114848) accepted by Expert Systems with Applications

by Qiangguo Jin, Hui Cui, Changming Sun, Zhaopeng Meng, Leyi Wei, Ran Su.

Example results

• Overview of the proposed DASC-Net for COVID-19 CT segmentation.

• Three examples of COVID-19 infection segmentation against Semi-Inf-Net. The segmentation results from Semi-Inf-Net were downloaded from the authors' GitHub repository(https://github.com/DengPingFan/Inf-Net) . The false predictions, i.e., false-positive and false-negative, are shown in red while the correct predictions are in green. The significant improvement by our model is marked with orange arrows.

• Visual comparison of COVID-19 infection segmentation against other methods on two target datasets.

• Illustration of segmentation map in 9 cycles on COVID-19-T1. The refinement is marked with orange arrows.

Dataset

CIR,ISMIR,MOSMEDDATA

Usage

• Releasing main code
• Detailed usage instruction

Citation

If the code is helpful for your research, please consider citing:

@article{JINDASC2021,
  title = "Domain adaptation based self-correction model for COVID-19 infection segmentation in CT images",
  author = "Qiangguo Jin, Hui Cui, Changming Sun, Zhaopeng Meng, Leyi Wei, and Ran Su",
  journal = "Expert Systems with Applications",
  year = "2021",
  pages = "114848"
}

Questions

General questions, please contact '[e-mail]'