Spectral prior image constrained compressed sensing (spectral PICCS) for photon-counting computed tomography

Area of Science:

• Medical Imaging
• Computational Imaging
• Photon-Counting Technology

Background:

• Photon-counting computed tomography (PCCT) enables multi-energy imaging from a single scan.
• PCCT divides the x-ray spectrum into narrower energy bins, increasing noise in each bin.
• Existing methods struggle with noise in these narrow energy bins.

Purpose of the Study:

• To develop an iterative reconstruction algorithm for noise suppression in PCCT narrow energy bins.
• To improve image quality in multi-energy PCCT applications.
• To address the noise challenge inherent in spectral data acquisition.

Main Methods:

• Proposed spectral PICCS (Prior Image Constrained Compressed Sensing) algorithm.
• Utilized full-spectrum image from filtered back-projection as a prior.
• Implemented using constrained optimization with adaptive iterative step sizes.
• Evaluated using simulations, physical phantoms, and in vivo swine studies.

Main Results:

• Substantial image noise reduction in narrow energy bins (43-73%).
• No sacrifice in CT number accuracy.
• No compromise in spatial resolution.
• Validated across simulations and experimental studies.

Conclusions:

• Spectral PICCS effectively suppresses noise in PCCT narrow energy bins.
• The algorithm maintains diagnostic image quality metrics.
• This technique enhances the utility of PCCT for multi-energy imaging.