Analytic simulation of petsird v0.7.2 data using parallelproj

This tool allows to "quickly" simulate and recostruct PETSIRD TOF listmode data using a simplified physics model (true coincidences only) and non-trivial detection efficiencies.

One of it's aims is to be able to generate data that can be used to test implementation of software supporting the PETSIRD standard for listmode data.

Create your conda / mamba environment

conda env create -f environment.yml
conda activate petsird-analytic-simulator
cd python

Simulate petsird LM data (work in progress)

python 01_analytic_petsird_lm_simulator.py

The simulation script creates a binary petsird LM file, but also many other files (e.g. a reference sensitivity image) that are all stored in the output directory

tree --charset=ascii data/sim_points_400000_0 

data/sim_points_400000_0
|-- ground_truth_image.npy
|-- reference_sensitivity_image.npy
|-- scanner_geometry.png
|-- sim_parameters.json
|-- simulated_petsird_lm_file.bin
|-- tof_profile_and_sensitivity_image.png
`-- zz.json

The simulation can be customized in many ways (number of counts to simulate, uniform or non-uniform efficiencies ...) via command line options.

These option can be listed via

python 01_analytic_petsird_lm_simulator.py -h

[-h] [--fname FNAME] [--output_dir OUTPUT_DIR] [--num_true_counts NUM_TRUE_COUNTS] [--skip_plots]
[--check_backprojection] [--num_epochs_mlem NUM_EPOCHS_MLEM] [--skip_writing] [--fwhm_mm FWHM_MM]
[--tof_fwhm_mm TOF_FWHM_MM] [--seed SEED] [--uniform_crystal_eff] [--uniform_sg_eff] [--img_shape IMG_SHAPE]
[--voxel_size VOXEL_SIZE] [--phantom {uniform_cylinder,squares,points}] [--num_time_blocks NUM_TIME_BLOCKS]
[--event_block_duration EVENT_BLOCK_DURATION]

options:
  -h, --help            show this help message and exit
  --fname FNAME
  --output_dir OUTPUT_DIR
  --num_true_counts NUM_TRUE_COUNTS
  --skip_plots
  --check_backprojection
  --num_epochs_mlem NUM_EPOCHS_MLEM
  --skip_writing
  --fwhm_mm FWHM_MM
  --tof_fwhm_mm TOF_FWHM_MM
  --seed SEED
  --uniform_crystal_eff
  --uniform_sg_eff
  --img_shape IMG_SHAPE
  --voxel_size VOXEL_SIZE
  --phantom {uniform_cylinder,squares,points}
  --num_time_blocks NUM_TIME_BLOCKS
  --event_block_duration EVENT_BLOCK_DURATION

Note: The "reference" MLEM using histogrammed data is only run if a value > 0 is given via --num_epochs_mlem. Otherwise it is skipped to save time.

Run a listmode OSEM recon on the simulated

python 02_reconstruct_petsird.py my_petsird_file.bin