Can virtual staining for High-throughput screening generalize?

Paper

Pytorch implementation of adapted pix2pixHD method for high-resolution (e.g. 1080x1080) virtual staining via image-to-image translation.

Prerequisites

• Linux or macOS
• Python 3
• NVIDIA GPU (11G memory or larger) + CUDA cuDNN

Getting Started

Create a new environment

conda create -n can_virtual_staining_for_high_thorughout_screening_generalize python=3.8
conda activate can_virtual_staining_for_high_thorughout_screening_generalize
pip install -e .

Clone this repo:

git clone [email protected]:krulllab/can_virtual_staining_for_high_thorughout_screening_generalize.git
cd src

Training

python ./src/train.py --dataroot ../path_to_data/ --data_type 16 --batchSize 4 --checkpoints_dir ../results/ --label_nc 0 --name experiment1 --no_instance  --resize_or_crop none --input_nc 1 --output_nc 1 --seed 42 --no_vgg_loss  --nThreads 1 --loadSize 256 --ndf 32 --norm instance --use_dropout  --fp16 --gpu_ids 1

• To view training results, please launch tensorboard --logdir opt.checpoints_dir

Multi-GPU training

python ./src/train.py --dataroot ../path_to_data/ --data_type 16 --batchSize 4 --checkpoints_dir ../../results --label_nc 0 --name experiment1 --no_instance  --resize_or_crop none --input_nc 1 --output_nc 1 --seed 42 --no_vgg_loss  --nThreads 1 --loadSize 256 --ndf 32 --norm instance --use_dropout  --fp16 --gpu_ids 1,2,3

Training with Automatic Mixed Precision (AMP) for faster speed

• To train with mixed precision support, please first install apex from: https://github.com/NVIDIA/apex
• You can then train the model by adding --fp16. For example,

python ./src/train.py --dataroot ../path_to_data/ --data_type 16 --batchSize 4 --checkpoints_dir ../results/experiment1/ --fp16

Testing

python test.py --results_dir ../results/inference/ --dataroot ../path_to_data/ --data_type 16 --batchSize 1 --checkpoints_dir ../results/experiment1/

The test results will be saved to a html file here: `./results/

Acknowledgments

This code borrows heavily from pytorch-CycleGAN-and-pix2pix and pix2pixHD