Sparse View CT Reconstruction with Projection Consistency Constraints
American Journal of Electrical Engineering and Technology · 31.01.2026 · 7 (1) · 95-101
DOI: 10.71465/ajeet3631 ↗Kurz gefasst
Fachartikel in der Zeitschrift American Journal of Electrical Engineering and Technology, online erschienen am 31.01.2026. Verfasst von James R. Walker und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Sparse-view CT reduces radiation exposure but increases streak artifacts due to insufficient projection angles. Motivated by convolution–self-attention denoising architectures such as CTLformer, this paper proposes a reconstruction framework that uses convolution–attention priors to regularize image updates while enforcing projection-domain consistency through an unrolled optimization scheme. The model alternates between a learned denoising block and a data-consistency step based on the forward projection operator. Experiments are conducted on two sparse-view CT benchmarks totaling 36,000 paired slices, with view counts ranging from 30 to 90. Comparisons against FBP, TV-regularized iterative reconstruction, RED-CNN post-processing, and transformerbased sparse-view models show PSNR gains of 1.2–2.0 dB and SSIM improvements of 0.015–0.028 under 60-view settings, with reduced streak intensity near high-contrast boundaries.
Zitieren
APA
Walker, J. R., Thompson, E. K., & Chen, D. M. (2026). Sparse View CT Reconstruction with Projection Consistency Constraints. American Journal of Electrical Engineering and Technology, 7(1), 95-101. https://doi.org/10.71465/ajeet3631
Vancouver
Walker JR, Thompson EK, Chen DM. Sparse View CT Reconstruction with Projection Consistency Constraints. American Journal of Electrical Engineering and Technology. 2026;7(1):95-101. doi:10.71465/ajeet3631
BibTeX
@article{Walker2026,
author = {Walker, James R. and Thompson, Emily K. and Chen, Daniel M.},
title = {Sparse View CT Reconstruction with Projection Consistency Constraints},
journal = {American Journal of Electrical Engineering and Technology},
year = {2026},
volume = {7},
number = {1},
pages = {95--101},
doi = {10.71465/ajeet3631}
}
EUVIMED