Skip to content

GitLab

GitLab's annual major release is around the corner. Along with a lot of new and exciting features, there will be a few breaking changes. Learn more here.

Last edited by AbedYassine
Page history

Home

What is PDT-SPACE?

PDT-SPACE is a software optimization tool to aid in interstitial photodynamic therapy (iPDT) planning. iPDT is a minimally invasive and less toxic (compared to conventional treatments) cancer therapy that relies on the light activation of photosensitizer drugs (PS), which leads, in the presence of Oxygen, to cytotoxic photo-chemical reactions that induce several cell death mechanisms such as apoptosis and necrosis. For an effective treatment, both the light dose and PS concentration should ideally be localized to the tumor region, so that damage to surrounding organs-at-risk (OAR) is minimized. For the light dose, clinicians follow certain treatment plans (protocols) that would result in *high-quality* treatments by deciding the number of light sources/diffusers to use, their positions, and the power allocated to each such that OAR damage is minimized. However, given that biological tissues are highly scattering media of light, and that the analytical model (radiative transfer equation) that represents this propagation is unsolvable for most realistic media, the problem of creating a high-quality, optimized plan is time-consuming and sometimes infeasible. Therefore, clinicians mostly resort to manually-created plans based on heuristics and previous attempts of iPDT. PDT-SPACE automates the process of iPDT planning by leveraging smart diffuser placement algorithms, convex optimization for power allocation, and a highly accurate and fast light propagation simulator named FullMonte. Additionally, PDT-SPACE generates robust plans by considering the statistical variability of optical properties (that govern how the light propagates) in the optimization framework, so that the risk of mis-predicting the treatment outcome is minimized.

Language

  • PDT-SPACE is written in C++11.
  • Lines of Code: approximately 15000

Overall flow

The overall flow of PDT-SPACE in its basic functionality is shown in the following flowchart:

flowchart TD
    
    subgraph Optimization
    F --> B((PDT-SPACE))
    B --> F((FullMonte))  
    end

    subgraph Inputs
    A(3D mesh of the<br>target tissues)  
    C(Tissue optical<br>properties)
    D(Tissue death PDT<br>dose thresholds) 
    E(Diffusers' initial<br>placement) 
    CC(Additional<br>constraints) 
    end

    subgraph Outputs
    G(Optimized<br>source powers)
    H(Optimized<br>diffuser placement)
    I(Final PDT<br>dose distribution)
    J(Dose-Volume<br>histograms)
    end
    Inputs --> Optimization
    Optimization-->Outputs

Installation Options

PDT-SPACE offers two options for building the project. You can use a docker container with either a precompiled image with PDT-SPACE already installed or an image where you need to build FullMonte from source with all dependencies already installed. The other option is a native install on your machine. This only works if you are running a UNIX distribution.

flowchart LR
    A[Install Options] --> B(Docker)
    A --> C(Native)
    subgraph "For Users"
      B --> D(Pre-Installed PDT-SPACE Image)
    end
    subgraph "For Developers"
      B --> E(PDT-SPACE source build Image)
      C --> F(Build PDT-SPACE from source)
    end
    click D "pdt-space-container" "Link to user install wiki"
    click E "Developer-Guide#docker-development-install" "Link to docker -dev install wiki"
    click F "Developer-Guide#native-development-install-linux-only" "Link to native install wiki"

For Users

For Users of PDT-SPACE, we recommend using our Docker image. Instructions on how to setup PDT-SPACE with Docker are given here:

For Developers

Developers have two options for installing PDT-SPACE:

  1. Docker installation
  • Recommended as all libraries and additional software is provided by the Docker image and does not need to be installed manually. Additionally, a pre-built version of PDT-SPACE is already available in the image which you can readily start developing in.
  1. Native installation
  • Some libraries must be installed natively to be able to develop in PDT-SPACE.
    • Vtk for visualization should be installed natively.
    • FullMonteSW for light propagation should be installed natively.
    • Mosek Fusion for optimization solver should be installed natively.

Citing the Tool

If you use results generated by PDT-SPACE in your publications, please reference at least one of the following papers:

@article{yassine2018automatic,
         title={Automatic interstitial photodynamic therapy planning via convex optimization},
         author={Yassine, Abdul-Amir and Kingsford, William and Xu, Yiwen and Cassidy, Jeffrey and Lilge, Lothar and Betz, Vaughn},
         journal={Biomed. Opt. Express},
         volume={9},
         number={2},
         pages={898--920},
         year={2018},
         publisher={Optical Society of America}
}
@article{yassine2019optimizing,
  title={Optimizing interstitial photodynamic therapy with custom cylindrical diffusers},
  author={Yassine, Abdul-Amir and Lilge, Lothar and Betz, Vaughn},
  journal={Journal of Biophotonics},
  volume={12},
  number={1},
  pages={e201800153},
  year={2019},
  publisher={Wiley Online Library}
}
@inproceedings{yassine2019tolerating,
  title={Tolerating uncertainty: photodynamic therapy planning with optical property variation},
  author={Yassine, Abdul-Amir and Lilge, Lothar and Betz, Vaughn},
  booktitle={Optical Methods for Tumor Treatment and Detection: Mechanisms and Techniques in Photodynamic Therapy XXVIII},
  volume={10860},
  pages={108600B},
  year={2019},
  organization={International Society for Optics and Photonics}
}