Anatomy-guided multi-resolution image reconstruction in PET.

Area of Science:

• Medical Imaging
• Computational Science

Background:

• Positron emission tomography (PET) imaging is crucial for medical diagnosis.
• Current reconstruction methods often face trade-offs between resolution, computational speed, and accuracy.

Purpose of the Study:

• To develop and evaluate a novel PET image reconstruction technique using a multi-resolution triangular mesh.
• To adapt the mesh resolution based on patient-specific anatomical data from CT or MRI.

Main Methods:

• A multi-resolution triangular mesh was adapted to patient-specific anatomy, with higher resolution in regions of interest (ROI).
• Maximum Likelihood Expectation Maximization (MLEM) reconstructions were performed using the multi-resolution mesh and compared against uniformly dense, sparse, and pixel-based meshes.
• Simulations used the NEMA image quality phantom and the XCAT human phantom.

Main Results:

• The multi-resolution mesh achieved reconstruction accuracy comparable to dense meshes in the ROI.
• The proposed method was computationally faster than uniformly dense mesh reconstructions.
• Reconstructions using the locally dense multi-resolution model were more accurate than pixel-based or sparse triangular meshes.

Conclusions:

• Patient-specific, multi-resolution triangular meshes offer a promising alternative to traditional pixel-based reconstruction in PET imaging.
• This approach balances high resolution in critical areas with computational efficiency.