Penalized-Likelihood PET Image Reconstruction Using 3D Structural Convolutional Sparse Coding.

Area of Science:

• Medical Imaging
• Image Reconstruction
• Computational Imaging

Background:

• Positron Emission Tomography (PET) is crucial for clinical diagnosis but limited by low resolution and high noise.
• Hybrid PET/CT and PET/MRI systems enable incorporating anatomical priors into PET reconstruction.
• Existing methods often require registration or supervised training, and patch-based approaches can introduce artifacts.

Purpose of the Study:

• To propose a novel penalized-likelihood PET image reconstruction method using cube-based 3D structural convolutional sparse coding (3D PET-CSC).
• To leverage anatomical priors effectively without registration or supervised learning.
• To improve PET image quality by reducing noise and artifacts.

Main Methods:

• Developed a cube-based 3D structural convolutional sparse coding (CSC) approach for PET image reconstruction (3D PET-CSC).
• Utilized convolutional operations to integrate anatomical priors directly into the 3D PET image.
• Incorporated residual-image and order-subset mechanisms to optimize computational efficiency and convergence speed.

Main Results:

• The proposed 3D PET-CSC method effectively incorporates anatomical priors, enhancing PET image quality.
• It alleviates staircase artifacts common in traditional patch-based sparse coding.
• Experimental results on simulated and clinical data show superior performance compared to reference methods.

Conclusions:

• 3D PET-CSC offers a significant advancement in PET image reconstruction, improving resolution and reducing noise.
• The method provides a straightforward and effective way to integrate anatomical information.
• This technique holds promise for enhancing diagnostic accuracy in clinical PET imaging.