Skip to content

Train VGG16 model and use CAM to generate pixel-level labels for CT scans.

Notifications You must be signed in to change notification settings

sino30535/Weak-label-in-CT-image-segamentation

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

26 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Weak label in CT image segamentation

Train models using using weak supervision such as image-level labels, producing pixel-level labels for CT scans.

Dataset

Folder hierarchy of dataset.

--Biomedical_data
  --train.py
  --predict.py
  --train256 
    --liver 
    --noliver 
  --test256 
    --images_pngs 
    --masks_pngs 

Perform transfer learning on CT scans dataset

python train.py --image_dir ./train256/
                --output_checkpoint ./output_model/checkpoint.ckpt
                --batch_size 16
                --epochs 10

Generate prediction for test256/images_pngs

python predict.py --image_dir ./test256/images_pngs/
                  --checkpoint_path ./output_model/checkpoint.ckpt
                  --output_dir ./output_prediction
                  --sensitivity 130

Result visualization

image

image

image

Reference

Learning Deep Features for Discriminative Localization http://cnnlocalization.csail.mit.edu/Zhou_Learning_Deep_Features_CVPR_2016_paper.pdf

Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization https://arxiv.org/pdf/1610.02391.pdf

About

Train VGG16 model and use CAM to generate pixel-level labels for CT scans.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published