Self-Adaptive Ring Artifact Reduction Method for Spectral Photon Counting CT.

Abstract

Photon Counting CT (PCCT) employs photon counting detectors (PCDs) which possess the unique ability to partition different energy ranges of the incoming photons. This fundamental function allows PCCT to differentiate materials in objects by fully utilizing the spectrum information through material decomposition. However, due to the differences in materials, fabrication and bonding, the spectrum response between each PCD pixel exhibits slight variation, leading to significant ring artifacts in direct reconstruction images. Furthermore, inaccurate material decomposition also induces ring artifacts in virtual monoenergetic image (VMI), which pose obstacles to clinical diagnosis. To address these issues and current bottle-necks of corresponding research area, we introduce a self-adaptive dual-domain fine-tuning strategy for ring artifact reduction. Specifically, we employ linear optimization on the projection-domain sinogram guided by edge-preserving image-domain constraints. This strategy efficiently detects and removes the ring artifacts on the reconstructed images while maintaining edge information. Visual and quantitative analyses on a clinical prototype system demonstrate the effectiveness of our ring artifact reduction method.