Skip to content

Latest commit

 

History

History
executable file
·
96 lines (51 loc) · 3.01 KB

README.md

File metadata and controls

executable file
·
96 lines (51 loc) · 3.01 KB

Machine-Learning-Applications-In-Digital-Pathology

A software system that automatically determines if a digital image of a pathology slide is cancerous The software system will be implemented in Python and use a variety of local and standard packages A user's guide for the software and an open source distribution that generic users can run"

How To Run:

Step 1: Add following to ~/.bashrc:

 . /data/isip/tools/GET_ENV.sh; # enable the isip conda environment
 
 MLADP=<path to github repo or whatever you renamed the github repo to > # path to the parent directory of the cloned repo IMPORTANT DO NOT HAVE BACKSLASH AT END
 	Ex: MLADP="/github_repo"
      
 PYTHONPATH="$NEDC_NFC/lib:$MLADP/nedc_mladp/lib:." # update Python path to contain requisite libraries
 
 export MLADP PYTHONPATH # export those two environment variables

Step 2 Activate new environment by executing in commandline:

 source ~/.bashrc

Step 3: Create a virtual environment by executing in comandline:

 python -m venv <name of virtual environment> --system-site-packages

Step 4: Activate virtual environment by executing in comandline:

 source <path of virtual environment>/bin/actiate

Step 5: Install required packages by executing in comandline:

 cat requirements.txt | xargs -n 1 pip install

 note:
You may get some errors here, but it should be okay!

Step 6 execute following commnds:

 cd $MLADP/nedc_mladp/src/

 ./make.sh

Step 7 create parameter files:

 cd $MLADP/nedc_mladp/data/

 In this directory these is a file called Example_Parameters.txt with an explanation of how to use the parameter file for our program.

Step 8 run the program:

 cd $MLADP/nedc_mladp/bin/

 ./nedc_mlad_run -p <absolute path to parameter file>

 I would recommend using this, it allso for no holdup and run in the background:

 nice -n 19 nohup ./nedc_mladp_run -p <absolute path to parameter file> &

 nice -n 19 nohup ./nedc_mladp_gen_feats -p <absolute path to parameter file> &

 nice -n 19 nohup ./nedc_mladp_train_model -p <absolute path to parameter file> &

 nice -n 19 nohup ./nedc_mladp_gen_preds -p <absolute path to parameter file> &

 nice -n 19 nohup ./nedc_mladp_gen_graphics -p <absolute path to parameter file> &

Notes:

If you clone this into an experiment directory into the TUH_dpath experiment folder and change the number from exp_0288 it should just run on the example files!

in order to extract a list of xml files and svs files, I recommend using the following commands.

list of xml files:

     find <absolute path of where your looking> -name *.xml>

     ex file:

     	parent/filepath1.xml
	parent/filepath2.xml
	parent/filepath3.xml

list of svs files:

     find <absolute path of where your looking> -name *.svs>

     ex file:

     	parent/filepath1.svs
	parent/filepath2.svs
	parent/filepath3.svs

at the time of this being written you can find a train, dev, and eval set here:

   /data/isip/data/tuh_dpath_breast/deidentified/v3.0.0/svs