Room: Room 202
Purpose: High requirement of tube voltage (kVp) switching frequency is the main challenge of rapid kVp-switching based multi-energy computed tomography (MECT) system. To address this problem, we propose an innovative sparse segmental MECT (SSMECT) scanning scheme, which is an easy-to-implement and no-extra dose induced way to realize on a conventional CT. A robust principle component analysis (RPCA) based method is also proposed to reconstruct SSMECT images.
Methods: For a sparse segmental scan, the x-ray source is controlled to maintain an energy within a segmental arc, and then switch alternately to another kVp level. Thus, we only need to switch tube kVp a few times to acquire multi-energy data. To solve the problem induced by this sparse and limited-angle acquisition, a RPCA-based reconstruction method is proposed, which includes three steps: 1) combine the sparse segmental projections into a reorganized dataset; 2) utilize filtered-back-projection (FBP) to reconstruct a prior image; 3) perform a RPCA-based iterative reconstruction with the spare segmental projections. Of which, the RPCA method considers the multi-energy image as the superposition of low rank and sparse components, fully utilizing the spectral correlation between different energy channels to remove noise and artifacts.
Results: A numerical simulation and a real phantom experiment were performed to demonstrate the efficacy and robustness of our scanning scheme and reconstruction method. The new scanning scheme was realized on a conventional CBCT platform. Visual inspection can consistently find that our proposed method suppress sparse-angle and limited-angle artifacts with good edge recovery. Quantitative evaluation shows that SSIM, MSE and CNR value of our method are better than those of some state-of-the-art methods.
Conclusion: Our study has demonstrated that this proposed SSMECT imaging approach could greatly lower kVp-switching frequency and achieve reasonable reconstruction accuracy and image quality.