CardiacLab

demo.mp4

Requirements

• miniconda (download here)
• 3D Slicer version 5.0 or later.
• 3D Slicer MONAI Label Plugin. (install step here)
• python 3.9

Install

conda create --name CardiacLab python=3.9 -y

pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 torchaudio==0.12.1 --extra-index-url https://download.pytorch.org/whl/cu113

pip install -r requirements.txt

Run Server

By cmd

monailabel start_server \
--app radiology \
--studies "D:\tmp\data" \
--conf models segmentation_cardiac \
--conf network unetcnx_a1 \
--conf --download_ckp_id <ID> \
--conf --target_spacing "0.7, 0.7, 1.0" \
--conf --spatial_size "128, 128, 128"\
--conf --intensity "-42, 423"

• --app : type of app.
• --studies: input data dir.
• --conf models: type of service.
• --conf network: type of network.
• --conf --download_ckp_id: download model checkpoint file from google drive.
• --conf --target_spacing: target spacing.
• --conf --spatial_size: spatial size (ROI).
• --conf --intensity: intensity.

By script

• set config to run/config.toml.

model = 'unetcnx_a1'

[segmentation_cardiac.unetcnx_a1]
app = "radiology"
studies = "D:\\tmp\\data"
models = "segmentation_cardiac"
network = "unetcnx_a1"
download_ckp_id = "<ID>"
target_spacing = "0.7, 0.7, 1.0"
spatial_size = "128, 128, 128"
intensity = "-42, 423"

• run run/segmentation.py.

python run/segmentation.py

Run Segmentation

• connect server

• upload data

• run

• done