Evaluation of Signal Enhancement Techniques for Motion Correction for High Temporal Resolution Functional PET

Area of Science:

• Neuroimaging
• Medical Physics
• Biomedical Engineering

Background:

• Functional Positron Emission Tomography (fPET) provides insights into brain energy metabolism and molecular connectivity.
• Advances in temporal resolution for fPET enable new research avenues.
• Short PET frames lead to lower signal-to-noise ratios (SNR), questioning standard motion correction efficacy.

Purpose of the Study:

• To evaluate standard motion correction for high temporal resolution fPET (3s frames).
• To assess if SNR enhancement methods improve motion correction accuracy.
• To compare deep learning, sliding window average, and filtering methods against conventional correction.

Main Methods:

• Investigated three SNR enhancement techniques: conditional generative adversarial networks (deep learning), exponentially weighted sliding window average, and a denoising filter.
• Compared motion correction performance by correlating motion parameters with fMRI data.
• Assessed task-induced activation magnitudes to evaluate correction effectiveness.

Main Results:

• None of the tested SNR enhancement methods significantly improved motion artifact mitigation.
• Standard motion correction without SNR enhancement proved adequate for the studied population.
• Advanced methods introduced increased computational cost without substantial benefits.

Conclusions:

• Standard motion correction is sufficient for high temporal resolution fPET in healthy subjects with typical movement levels.
• Current SNR enhancement techniques do not offer significant advantages for motion correction in this context.
• Future research should focus on developing targeted strategies to further enhance motion correction in fPET.