unmixing_pipeline:

unmixing_pipeline is a tool for unmixing Immunohistochemistry (IHC) images taken in multiple channels and rounds.

Dependencies:

• numpy
• scipy
• scikit-image
• scikit-learn
• tifffile (conda-forge)
• matplotlib

Pipeline:

You can run the unmxing pipeline in 2 modes:

1. main_unsupervised: In this mode, all the channels of same round are going to be unmixed with sparse spectral unmixing algorithm.

   python main_unsupervised.py --img_dir=/path/to/input/dir \
                               --save_dir=/path/to/save/dir \
                               --round_pattern=R \
                               --channel_pattern=C \
                               --has_brightfield=True    

   It will save unmixed images and generate a script of unmixed channels in the save_dir:

2. main_supervised: In this mode, user can correct the script and all the channels provided in the script are going to be unmixed with chemical co-localization unmixing algorithm.

   

   python main_supervised.py --img_dir=/path/to/input/dir \
                             --save_dir=/path/to/save/dir \
                             --script_file=/path/to/script/file   

   It will save unmixed images in save_dir.

Arguments:

1. main_unsupervised:

   Argument	Help	Example
   img_dir	Path to the directory of images	--img_dir=C:\images\input
   save_dir	Path to the directory to save unmixed images	--save_dir=C:\images\output
   default_box	Selected box coordinates xmin_ymin_xmax_ymax	--default_box=16200_6100_21300_12200
   has_brightfield	If last channel is brightfield	--has_brightfield=True
   round_pattern	Pattern for round idx	--round_pattern=R
   channel_pattern	Pattern for channel idx	--channel_pattern=C

2. main_supervised:

   Argument	Help	Example
   img_dir	path to the directory of images	--img_dir=C:\images\input
   save_dir	path to the directory to save unmixed images	--save_dir=C:\images\output
   script_file	csv script file name	--script_file=script.csv
   default_box	selected box coordinates xmin_ymin_xmax_ymax	--default_box=16200_6100_21300_12200
   visualize	visualize the unmixing report of crop	--visualize

Notes:

• if --visualized passed: