Inversion of the 3D Radon transform for a multidetector, point-focused SPECT brain scanner

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Image Reconstruction

Background:

• The Radon transform is fundamental for image reconstruction in emission tomography.
• Traditional 2D reconstruction methods in SPECT may introduce artifacts, particularly for non-cylindrical sources.
• Point-focused SPECT scanners require precise reconstruction algorithms to accurately represent brain structures.

Purpose of the Study:

• To evaluate the accuracy of a 3D Radon transform reconstruction for unattenuated SPECT data.
• To compare the performance of 3D and 2D reconstruction methods for simulated brain SPECT data.
• To assess the impact of reconstruction method on image fidelity for disc-shaped objects.

Main Methods:

• Application of the 3D Radon transform to simulated unattenuated projection data from a multidetector SPECT scanner.
• Comparison with the conventional 2D ramp filter reconstruction algorithm.
• Computer simulation of projection data for cylindrical and disc-shaped sources with identical parameters.

Main Results:

• The 3D reconstruction method achieved over 97% accuracy for the central slice of a disc-shaped object.
• The conventional 2D method resulted in a significant (20%) central dip artifact in the disc reconstruction.
• The 3D method exhibited a 10.1-fold lower noise-equivalent sensitivity for the disc compared to the 2D method for a cylindrical source.

Conclusions:

• The 3D Radon transform is essential for accurate reconstruction of SPECT brain data, especially for non-ideal geometries.
• The 2D ramp filter method is a limiting case of the 3D method and is only exact for sources with cylindrical symmetry.
• Implementing the 3D method improves image quality and reduces artifacts in SPECT brain imaging.