VascuSynth-AI

VascuSynth - Vascular Tree Synthesis Software Originally software by Preet S. Jassi, May 5 2011

Please visit the Insight Journal and read the paper here http://www.insight-journal.org/browse/publication/794 for proper instructions on how to compile and execute VascuSynth. VascuSynth requires CMake and ITK. Instructions on how to install CMake and ITK are in the paper above.

Motivation

This repository is a modified version of VascuSynth to generate 3D vascular trees with tumourous regions. The goal is to generate realistic MIPs of 3D vasculature trees with hypoxic regions that reflect key features such as tortuosity, dilated vessels, and dense branching in contrast to the organized nature of healthy vasculature. Future work involves generated large datasets to evaluate the ability of AI networks to identify and segment tumourous vasculature from these images (intended to simulate real-world ultrasound acquisition).

Build and Generate Vascular Trees

This application can be built and run using Docker.

It will first prompt the user to enter parameters such as # of trees, # of nodes, and whether they would like to generate tumourous or healthy vasculature. After the configuration files are built, the program will ask the user if they would like to generate the vascular trees. The container exits after the trees have been generated (or if the user decides not to generate them at all)

First, clone the repo:

git clone https://github.com/jkaethee/VascuSynth_copy.git

Then, start Docker and build the image:

docker build -t vascusynth .

Finally, spin up a container in interactive mode to run the application:

docker run -it vascusynth

The user can now follow the interactive prompts and generate their desired vascular trees!

Tumourous vascular trees can be found in the /VascuSynth/Tumourous_Trees directory, while healthy vascular trees can be found in the /VascuSynth/Healthy_Trees directory (both of these paths are on the container file system)

Copying files from container to host

To get the files generated within the container onto the user's host system, use the following command:

docker cp <container_name>:/VascuSynth/<Tumourous_Trees or Healthy_Trees> ./

Note that the container name can be found by running docker ps -a.
If successful, the user should see the folders and files corresponding to the generated trees on their host system.

Capture Maximum Intensity Projections of 3D Vascular Trees

This section utilizes Jupyter Notebook. To install Jupyter Notebook, run the following command in your terminal:

pip install notebook

Then, navigate to the Source directory and run:

jupyter notebook

This should run the the notebook and allow you to view and run the MIP_Generator.ipynb file. This notebook uses 3D slicer to view, rotate, and capture MIP images for every 10 degrees of rotation in the pitch, yaw, and roll axis. Ensure that the folder directories and names correspond with those in your VascuSynth folder before running the cells in the notebook.