IHC Analysis Framework
_______________
IHC Analysis Framework is a package to locate and label different biomarkers on multi-stained immunohistochemistry (IHC) images:
Please see the paper:
"Automatic labeling of molecular biomarkers of imunohistochemistry images using fully convolutional networks". Fahime Sheikhzadeh, Rabab K. Ward, Dirck van Niekerk, Martial Guillaud
data provided here can replicate the exact results reported in the paper. The framework is easy to extend for other applications.
Dependencies
_______________
* Caffe deep learning framework
* Python 2.7 (to run pycaffe)
* Matlab
All the analysis has been done on linux.
Replication of the results reported in the paper
_________________________________________________
The data related to developed N-Net, including Nuclei Images used as train set and test set, and N-Net parameters, are provided in the folder "Nuclei Network".
The data related to developed WI-Net, including Whole IHC Images, and WI-Net parameters, are provided in the folder "Whole Image Network".
The data related to proposed IHC Layer Analysis approach, are located in the folder "Layer Analysis". To replicate the graphs in the paper, download "Labeled_IHC_Images" folder, and run "IHCLayerAnalysis.m". This approach could be extended to other datasets, if labeled images and ROIs are located in the same directory formats.
____________________________________________________________________________________________________________
If you have any questions about this software please feel free to contact 'fahime@ece.ubc.ca'