• Quick summary
  • A Python-based Dynamic multi-bed PET simulation and reconstruction framework with analytical modeling of system matrix, capable of incorporating various degrees of PSF modeling.
• Version
  • 1.0.1.1

• Necessary packages

  • NiBabel (To read and write NIfTI files)
  • scikit-image
  • SciPy
  • sklearn

• Start-up

  • To start the project you need to run the file SimulateDynamicMultibedFDG.py

• Configurations (config.json)

  • You can change various parameters in the config file such as:
    • Dimensions (xdim, ydim, zdim)
    • Number of frames
    • Axial and trans axial Field of View
    • Iterations
    • Subsets
    • Scan Duration
    • VCT Sensitivity
    • Attenuation units (if mu_map is in units of 1/voxel attenuation, that's the default assumption. If it's in units of 1/mm, then one needs to multiply it by bins-size)
  • Additional configurations can be added to automate and ease the process of simulation.

• Image Comparison

  • File CompareOutputImages.py calculates the following metrics between MATLAB and Python generated files:
    • Mean Squared Error (MSE)
    • Peak Signal to Noise Ratio (PSNR)
    • Structural Similarity Index (SSIM)
  • This file is run independently.

This code was developed by Nolla Sherifi ([email protected]) between October 2023 and March 2024 at the Technical University of Munich.

• Directory structure (cleaned up i/o)
• Removed some input parameters (dimension automatically grabbed from image, etc)
• Fixed bug in bins calculation (previously not working for image dimensions other than 128)
• New input for mu map units ("/cm", "/mm", or "/voxel")
• Added new full length xcat phantom
• SimulateDynamicMultibedFDG.py is now main.py. Run this script to start simulation.
• tqdm (progress bars) was added, so this is a newly required dependecy (or it can be removed fairly easily)