Low-noise dynamic reconstruction for x-ray tomographic perfusion studies using low sampling rates.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Radiology

Background:

• Functional imaging using X-ray techniques requires high temporal resolution and low radiation dose per scan.
• Current methods often compromise signal-to-noise ratio (SNR) due to fast data acquisition and low-dose scans.
• Previous work introduced temporal interpolation to address temporal resolution constraints.

Purpose of the Study:

• To improve the signal-to-noise ratio (SNR) in functional X-ray imaging.
• To develop a method that reduces noise dependence on the number of scans and individual scan SNR.
• To enable perfusion imaging with slowly rotating CT systems.

Main Methods:

• Implemented a temporal interpolation scheme in the projection data space.
• Introduced a temporal smoothing term within the interpolation scheme.
• Filtered data to consider only physiologically relevant bandwidths.

Main Results:

• Achieved a significant increase in the signal-to-noise ratio (SNR).
• The noise level in reconstructed images became dependent only on the total applied radiation dose.
• Noise independence from the number of scans and single-scan SNR was demonstrated.

Conclusions:

• The enhanced temporal interpolation scheme significantly boosts SNR in functional X-ray imaging.
• This approach decouples noise from scan number and individual scan quality, relying solely on total dose.
• The method represents a potential advancement for perfusion imaging using C-arm and small animal CT scanners.