Premium
WE‐G‐18A‐05: Cone‐Beam CT Reconstruction with Deformed Prior Image
Author(s) -
Zhang H,
Ouyang L,
Huang J,
Ma J,
Chen W,
Wang J
Publication year - 2014
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4889516
Subject(s) - artificial intelligence , computer vision , cone beam computed tomography , projection (relational algebra) , imaging phantom , iterative reconstruction , image quality , image (mathematics) , computer science , image registration , image restoration , image guided radiation therapy , medical imaging , mathematics , image processing , algorithm , nuclear medicine , computed tomography , medicine , radiology
Purpose: Prior image can be incorporated into image reconstruction process to improve the quality of on‐treatment cone‐beam CT (CBCT) from sparseview or low‐dose projections. However, the deformation between the prior image and on‐treatment CBCT are not considered in current prior image based reconstructions (e.g., prior image constrained compressed sensing (PICCS)). The purpose of this work is to develop a deformed‐prior‐imagebased‐ reconstruction strategy (DPIR) to address the mismatch problem between the prior image and target image. Methods: The deformed prior image is obtained by a projection based registration approach. Specifically, the deformation vector fields (DVF) used to deform the prior image is estimated through matching the forward projection of the prior image and the measured on‐treatment projection. The deformed prior image is then used as the prior image in the standard PICCS algorithm. Simulation studies on the XCAT phantom was conducted to evaluate the performance of the projection based registration procedure and the proposed DPIR strategy. Results: The deformed prior image matches the geometry of on‐treatment CBCT closer as compared to the original prior image. Using the deformed prior image, the quality of the image reconstructed by DPIR from few‐view projection data is greatly improved as compared to the standard PICCS algorithm. The relative image reconstruction error is reduced to 11.13% in the proposed DPIR from 17.57% in the original PICCS. Conclusion: The proposed DPIR approach can solve the mismatch problem between the prior image and target image, which overcomes the limitation of the original PICCS algorithm for CBCT reconstruction from sparse‐view or low‐dose projections.