WE-C-217BCD-09: Compressed Sensing in PET Imaging with Partial Detector Rings

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Signal Processing

Background:

• Positron Emission Tomography (PET) systems typically require a high density of detector elements for optimal image reconstruction.
• Reducing the number of detectors per ring could decrease hardware costs and system complexity.
• Compressed sensing (CS) offers a potential solution for reconstructing images from undersampled data.

Purpose of the Study:

• To investigate the feasibility of reducing PET detector elements per ring by introducing gaps.
• To evaluate if compressed sensing techniques can maintain PET image quality despite detector reduction.

Main Methods:

• A Ge-68 phantom was imaged on a D-RX PET/CT scanner in a baseline (full detectors) and partially sampled (PS) mode (11% detectors off).
• A compressed sensing algorithm iteratively reconstructed the PS sinogram by decomposing it into sparse components in different transform domains.
• Reconstructed images were compared to baseline images using quantitative metrics (mean and max activity concentration).

Main Results:

• Reconstructed PET images using OSEM and FBP showed minimal differences compared to baseline.
• Mean activity concentration differences were -0.06% (OSEM) and -0.02% (FBP).
• Max activity concentration differences were 3.7% (OSEM) and 5.5% (FBP).

Conclusions:

• Compressed sensing effectively recovers data from partially sampled PET scans.
• This approach holds promise for developing PET systems with fewer detectors per ring.
• Accurate quantitation can potentially be achieved with reduced detector elements, lowering costs and complexity.