DocumentCode
3342098
Title
Iterative motion-aware reconstruction algorithm for cardiac cone beam CT
Author
Schomberg, Hermann
Author_Institution
Philips Res., Hamburg, Germany
fYear
2011
fDate
23-29 Oct. 2011
Firstpage
4077
Lastpage
4079
Abstract
It is feasible, though challenging, to build a cone beam CT scanner with a rotating gantry that can make a full turn in about 270 ms and has a large area detector mounted to it. Given such a scanner, one might try to reconstruct a 3-D still image of the beating heart from a single circular short scan of the thorax. Such a scan would take about 150 ms, which is still appreciably longer than the resting phase of the heart. So even if the resting phase fell within the time window of the data acquisition, any reconstruction algorithm for static cone beam CT would produce an image still somewhat affected by motion artifacts. In this paper, it is proposed to avoid such artifacts by means of a motion-aware extension of the Algebraic Reconstruction Technique. The proposed algorithm solves a mathematical model of the data acquisition that allows the heart to move during the scan. The motion of the heart is represented by a vector field which is reconstructed jointly with the image and need not be known beforehand.
Keywords
cardiology; computerised tomography; data acquisition; diagnostic radiography; image reconstruction; image scanners; iterative methods; medical image processing; 3-D still image reconstruction; algebraic reconstruction technique; beating heart; cardiac cone beam computerised tomography; data acquisition; iterative motion-aware reconstruction algorithm; large area detector; motion artifacts; motion-aware extension; resting phase; rotating gantry; single circular short scan; thorax; Image reconstruction;
fLanguage
English
Publisher
ieee
Conference_Titel
Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC), 2011 IEEE
Conference_Location
Valencia
ISSN
1082-3654
Print_ISBN
978-1-4673-0118-3
Type
conf
DOI
10.1109/NSSMIC.2011.6153775
Filename
6153775
Link To Document