TY - GEN
T1 - Prior-Based Enhanced ASD-POCS for Artifact Suppression and Structural Preservation in Sparse-View CBCT
AU - Bappy, D. M.
AU - Kang, Donghwa
AU - Lee, Jinkyu
AU - Baek, Hyeongboo
N1 - Publisher Copyright:
© 2024 Copyright is held by the owner/author(s). Publication rights licensed to ACM.
PY - 2024/4/8
Y1 - 2024/4/8
N2 - Among CT (Computed Tomography) techniques that produce cross-section images by acquiring X-ray data from multiple angles around the individual, CBCT (Cone Beam CT) that collects the data through cone beam is popular due to its potential to reduce radiation risks. Although employing low-dose and sparse-view CBCT is a cornerstone approach to reducing radiation risks, reconstructing CBCT from noisy and sparsely-acquired cone beam scans often result in significant artifacts due to ill-posed inverse problem.In this paper, we aim to suppress the artifacts derived from reconstructing CBCT. To this end, we target the well-known existing approach ASD-POCS (Adaptive Steepest Descent with Projections Onto Convex Sets), focus on its downside of potentially introducing over-smoothing near edges and removing intricate fine structures, and develop an advanced ASD-POCS strategy that addresses the downside by leveraging prior image information. Our approach to integrating the prior information not only maintains structural integrity (without compromising edge detail or erasing fine structures) but also diminishes artifacts, thereby counteracting noise and blur to a significant extent. Our solution underwent stringent testing using both simulated data and publicly available datasets. The testing results demonstrate that the proposed technique is robust to the prevailing challenges in sparse-view CBCT reconstruction, skillfully mitigating artifacts (compared to contemporary state-of-the-art methods) while safeguarding intricate structures.
AB - Among CT (Computed Tomography) techniques that produce cross-section images by acquiring X-ray data from multiple angles around the individual, CBCT (Cone Beam CT) that collects the data through cone beam is popular due to its potential to reduce radiation risks. Although employing low-dose and sparse-view CBCT is a cornerstone approach to reducing radiation risks, reconstructing CBCT from noisy and sparsely-acquired cone beam scans often result in significant artifacts due to ill-posed inverse problem.In this paper, we aim to suppress the artifacts derived from reconstructing CBCT. To this end, we target the well-known existing approach ASD-POCS (Adaptive Steepest Descent with Projections Onto Convex Sets), focus on its downside of potentially introducing over-smoothing near edges and removing intricate fine structures, and develop an advanced ASD-POCS strategy that addresses the downside by leveraging prior image information. Our approach to integrating the prior information not only maintains structural integrity (without compromising edge detail or erasing fine structures) but also diminishes artifacts, thereby counteracting noise and blur to a significant extent. Our solution underwent stringent testing using both simulated data and publicly available datasets. The testing results demonstrate that the proposed technique is robust to the prevailing challenges in sparse-view CBCT reconstruction, skillfully mitigating artifacts (compared to contemporary state-of-the-art methods) while safeguarding intricate structures.
KW - artifacts
KW - computed tomography
KW - FISTA
KW - low-dose
KW - sparse-view
UR - http://www.scopus.com/inward/record.url?scp=85197689234&partnerID=8YFLogxK
U2 - 10.1145/3605098.3635910
DO - 10.1145/3605098.3635910
M3 - Conference contribution
AN - SCOPUS:85197689234
T3 - Proceedings of the ACM Symposium on Applied Computing
SP - 496
EP - 505
BT - 39th Annual ACM Symposium on Applied Computing, SAC 2024
PB - Association for Computing Machinery
T2 - 39th Annual ACM Symposium on Applied Computing, SAC 2024
Y2 - 8 April 2024 through 12 April 2024
ER -