Proton Beam Therapy Reinforcement Learning

Overview

This repository provides a framework for setting up reinforcement learning (RL) on patient data to create personalized treatment plans using Proton Beam Therapy (PBT). The current implementation includes data processing and beam modeling components essential for simulating and analyzing proton beam interactions with patient-specific density maps.

Table of Contents

• Features
• Requirements
• Installation
• Usage
  • Data Processing
  • Running Beam Models
• File Structure
• Contributing
• License
• Quick Start Example
• Contact

Features

• Data Processing: Converts and prepares DICOM data for use in beam modeling.
• Beam Modeling: Simulates proton beam interactions with patient-specific density maps and visualizes dose distributions.
• Extensible Framework: Designed to incorporate reinforcement learning algorithms for optimizing treatment plans.

Requirements

• Operating System: Windows
• Python Version: 3.8 or higher
• Libraries:
  • numpy
  • matplotlib
  • pydicom
  • opencv-python
  • scipy
  • pickle
  • protonbeam (custom module)

Installation

1. Clone the Repository

   git clone https://github.com/haydentbs/PBT_RL_GAMIFICATION.git
   cd PBT_RL_GAMIFICATION

2. Set Up a Virtual Environment (Optional but Recommended)

   python -m venv venv
   venv\Scripts\activate

3. Install Dependencies

   pip install -r requirements.txt

   If requirements.txt is not available, you can install the necessary libraries manually:

   pip install numpy matplotlib pydicom opencv-python scipy

   Ensure that the custom ProtonBeam module is available in your Python path.

Usage

Data Processing

Before running the beam simulation, you need to process the patient DICOM data.

1. Prepare DICOM Data

   Ensure that your DICOM files are placed in the appropriate directory as specified in data_processing/data_processing.py. By default, the base path is set to:

   base_path = r"data"

2. Run Data Processing Script

   Execute the data_processing/data_processing.py script to process the DICOM data and generate the necessary pickle files.

   python data_processing/data_processing.py

   This will create processed data files, such as:

   • pickles/example_dicom_processed_data.pkl

Running Beam Models

After processing the data, you can run the beam simulation to generate and visualize proton beam dose distributions.

1. Run Beam Test Script

   Execute the test_beam.py script to simulate the proton beam and visualize the results.

   python test_beam.py

   This script performs the following:

   • Loads the processed DICOM and beam data.
   • Simulates a proton beam at specified coordinates and angle.
   • Visualizes the dose distribution on the density map.
   • Generates dose vs. water equivalent distance plots.

File Structure

   

proton-beam-therapy-rl/
│
├── data_processing/
│ └── data_processing.py
│
├── pickles/
│ ├── example_dicom_processed_data.pkl
│ └── beam_data.pkl
│
├── structures_to_display.py
│
├── test_beam.py
│
├── ProtonBeam/
│ ├── init.py
│ └── ProtonBeamV4.py
│
├── requirements.txt
│
└── README.md

• data_processing/: Contains the data processing scripts.
• pickles/: Stores processed data files.
• structures_to_display.py: Defines structures for display purposes.
• test_beam.py: Script for beam simulation and visualization.
• ProtonBeam/: Custom module for proton beam modeling.
• requirements.txt: Lists Python dependencies.
• README.md: Project documentation.

License

This project is licensed under the MIT License.

Quick Start Example

1. Run Data Processing

   python data_processing/data_processing.py

2. Run Beam Simulation

   python test_beam.py

3. View Generated Plots

   The script will display two plots:

   • Dose vs. water equivalent distance plot for the phantom.
   • Dose beam through a patient CT slice.

   

   

Ensure that all dependencies are installed and the data paths in the scripts are correctly set according to your system.

For further information, please contact hayden@htibbals.com.