Nucleui segmentation

This source includes nucleui segmentation code using deep learing framework(caffe).
We converted matlab code[2] to python code to run without MATLAB and implemented on ubuntu 16.04, python3.5, and caffe.

Requirments

python3.5
caffe

Overview

Data

Download datasets

   tar -xvzf nuclei.tgz
   mv nuclei/*  ~/public/DL_tutorial_Code/1-nuclei/images

Pre-trained model and prototxt files

   cd DL_tutorial_code/1-nuclei/models/
   cd DL_tutorial_code/common/

Training/Testing

   cd tuturial_py/
   step1_patch_extraction.py 
   step2_cross_validation_creation.py (training and testing list creation step)
   step3_make_db.py (database creation step)
   step4_submit_jobs.py (training step)
   step5_create_output_images_kfold.py (testing step)

Acknowledgements

We would like to thank the authors of DLtutorialCode[2], which we use in this work.

References

[1]Janowczyk, A., Madabhushi, A., 2016. Deep learning for digital pathology image analysis: A comprehensive tutorial with selected use cases. Journal of Pathology Informatics 7, 29.
[2]original source