Skip to content

A 3D CNN Network with BERT For Automatic COVID-19 Diagnosis From CT-Scan Images (https://arxiv.org/abs/2106.14403)

Notifications You must be signed in to change notification settings

deepcam-cn/3D-CNN-BERT-COVID19

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

30 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

3D-CNN-BERT-COVID19: The 4th-place solution for the ICCV-2021 MIA COV19D Competition.

Implementation of "A 3D CNN Network with BERT For Automatic COVID-19 Diagnosis From CT-Scan Images" For details, refer to our paper https://arxiv.org/abs/2106.14403

There are four parts in this project

Preprocess

Preprocess the CT-scan volume images: check the image size, extract bounding box and percentage of the the lung in the whole image, select images for 3D CNN

Segmentation

A UNet segmentation network is trained. It is used to segment lung mask of an image.

BERT

A 3D CNN network with BERT for CT-scan volume classification and embedding feature extraction

MLP

A simple MLP is trained on the extracted 3D CNN-BERT features. This helps the classification accuracy when there are more than one set of images in a CT-scan volume.

License

The code of 3D-CNN-BERT-COVID19 is released under the MIT License. There is no limitation for both academic and commercial usage.

About

A 3D CNN Network with BERT For Automatic COVID-19 Diagnosis From CT-Scan Images (https://arxiv.org/abs/2106.14403)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published