/pdusnet

fCNN framework for 3D ultrasound segmentation using B-Mode and/or other modality data

Primary LanguagePython

Framework for segmenting 3D Ultrasound

Getting Started

Please refer to requirements.txt for all the packages that need to be installed.

Run the following commands to get the environment setup.

export TF_ENABLE_AUTO_MIXED_PRECISION=1 && export PATH=$PATH:/usr/local/cuda-10.0/bin && export CUDADIR=/usr/local/cuda-10.0 && export CUDA_VISIBLE_DEVICES=0,1 && export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-10.0/lib64 && export TF_FORCE_GPU_ALLOW_GROWTH=true

Running network

Update config.py with any configuration settings you wish to change.

Create datasets

python write.py --augment=True

# Non Augmented
python write.py

Train and evaluate model

# Single modal
python train.py --model=unet --batch_size=10 --num_epochs=50
python train.py --model=unet++ --batch_size=10 --num_epochs=50

# Early Fusion
python train.py --model=unet --batch_size=10 --num_epochs=50 --multi_modal=True --early_fusion=True
python train.py --model=unet++ --batch_size=10 --num_epochs=50 --multi_modal=True --early_fusion=True

# Layer Fusion
python train.py --model=unet --batch_size=10 --num_epochs=50 --multi_modal=True
python train.py --model=unet++ --batch_size=10 --num_epochs=50 --multi_modal=True

# Late Fusion
python train.py --model=unet --batch_size=10 --num_epochs=50 --multi_modal=True --late_fusion=True
python train.py --model=unet++ --batch_size=10 --num_epochs=50 --multi_modal=True --late_fusion=True