Dynamic PET simulator via tomographic emission projection for kinetic modeling and parametric image studies

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Computational Science

Background:

• Dynamic Positron Emission Tomography (PET) simulations are crucial for research and education.
• Monte Carlo (MC) simulations offer high accuracy but are computationally intensive.
• There is a need for faster, simpler simulation tools for dynamic PET analysis.

Purpose of the Study:

• To develop and evaluate dpetstep (Dynamic PET Simulator of Tracers via Emission Projection), a fast and simple tool for dynamic PET simulations.
• To provide an alternative to MC simulations for educational purposes and evaluating clinical environment effects.
• To assess the impact of postprocessing choices on dynamic and parametric PET images.

Main Methods:

• Developed dpetstep in MATLAB, integrating existing and new modules.
• Simulated time-activity curves for each voxel, incorporating system blurring, noise, scatter, randoms, and attenuation.
• Reconstructed simulated frames using user-specified methods and compared results with MC data and Gaussian-noised curves (GAUSS).

Main Results:

• dpetstep was 8000 times faster than MC simulations.
• Dynamic images from dpetstep showed a 4% average root mean square error compared to MC, versus 11% for GAUSS.
• Noise profiles in dpetstep images closely matched MC, with no significant statistical differences in tumor regions (p < 0.01).

Conclusions:

• dpetstep provides a fast, easy, one-stop solution for dynamic PET and parametric image simulations.
• The tool generates images with noise properties highly similar to MC images in a fraction of the time.
• While useful for general simulations, dpetstep's simplified scatter and random models may limit its suitability for specific phenomenon investigations.