OMEGA-m

OMEGA-m reconstructs positron emission tomography (PET) images and facilitates development and evaluation of PET reconstruction algorithms.

Key Features:

• MATLAB/GNU Octave integration: Implemented for use within MATLAB and GNU Octave environments for scripting and algorithm development.
• Matrix-free forward/back projections: Provides matrix-free implementations of forward (A * x) and backward (A' * y) projections for reconstruction and algorithm testing.
• GATE simulated data support: Directly supports input from GATE Monte Carlo simulations for algorithm evaluation with simulated PET data.
• Parallel computing: Employs OpenMP for CPU parallelization and OpenCL for GPU acceleration.
• Correction functions: Includes normalization correction and supports attenuation, randoms, and scatter corrections.
• Algorithmic implementations: Implements maximum-likelihood and maximum a posteriori (MAP) algorithms with various priors and accepts user-supplied priors.
• System matrix options: Allows reconstruction using explicitly computed system matrices or a matrix-free formalism, with OpenCL acceleration for matrix-free operations.

Scientific Applications:

• Algorithm testing and development: Enables development and benchmarking of new PET image reconstruction algorithms using implemented ML and MAP methods.
• Simulated data analysis: Supports analysis and validation of reconstruction approaches using GATE Monte Carlo simulated PET datasets.
• Clinical research: Provides reconstruction and correction functionalities applicable to PET imaging studies in clinical research contexts.

Methodology:

Implemented in MATLAB and GNU Octave; supports matrix-free forward (A * x) and backward (A' * y) projections or explicitly computed system matrices; implements maximum-likelihood and MAP algorithms with user-specified priors; uses OpenMP for CPU and OpenCL for GPU acceleration; includes normalization, attenuation, randoms, and scatter corrections; accepts GATE simulated data.