PAGE-Net

A biologically interpretable integrative deep learning model that integrates PAthological images and GEnomic data

PAGE-Net has three phrases:

• Patch-wise pre-trained CNN
• Two-stage Aggeragation
• Integration of aggregated pathological images and genomic data

Get Started

Patch-wise pre-trained CNN (see pretrain folder)

1. patch_extraction : it extracts valid patchs from WSI by removing background and stains.
2. Datagenerator.py : it makes the data ready for Keras image data loaders. (As data is large we used keras dataloaders for loading data)
3. PAGE_net_pretrain : code for pretraining and saving the pretrained model.

Two-stage aggregation (see aggregation folder)

1. Aggergation.py : it generates the aggregated score and saves in csv.
2. DataMatching.py : it splits the aggregated data in train, test, and validation

Integration of aggregated pathological images and genomic data

Run.py: to train the model with the inputs from train.csv. Hyperparmeters are optimized by grid search automatically with validation.csv. C-index is used to evaluate the model performance with test.csv.

#Package Requirments ###Preprocessing

• PyHistopathology package
• Keras
• Tensorflow
• Keract