Automatic motion correction in cone-beam computed tomography

S. Ens, J. Ulrici, E. Hell, T. M. Buzug


Motion correction is an important task in computed tomography. Small movements lead to a reduced resolution of the reconstructed images; strong movements may lead to a false diagnosis. Thus, motion detection and subsequent correction are necessary. In this work, we propose an algorithm for the automatic estimation of movement positions and motion correction by the simultaneous estimation of motion parameters and the determination of the corrected reconstruction. Distance metric values, computed from two successive projection images, provide information of the incorporation of movement. Reconstruction correction is achieved by minimization of an image quality function. Two no-reference image metrics show sufficient correlation with the amount of motion artifacts: entropy and bandpass-metric. The aim of the optimization method is to find a pair of image and motion parameters by minimization of the metric. However, the described method for the estimation of movement positions can also be used as a pre-processing step of the data-driven motion-correction (DDMC), a recently published motion correction method, for subdivision of the projection data into motion free subsections. Since the proposed method uses the Feldkamp algorithm (FDK) for reconstruction, we expect a shorter computation time then for the iterative reconstruction based DDMC. The proposed method works solely on information contained in cone-beam projections.

Original languageEnglish
Title of host publicationIEEE Nuclear Science Symposuim & Medical Imaging Conference
Number of pages4
Publication date01.12.2010
Article number5874405
ISBN (Print)978-1-4244-9106-3, 978-1-4244-9104-9
ISBN (Electronic)978-1-4244-9105-6
Publication statusPublished - 01.12.2010
Event2010 IEEE Nuclear Science Symposium, Medical Imaging Conference - Knoxville, United States
Duration: 30.10.201006.11.2010


Dive into the research topics of 'Automatic motion correction in cone-beam computed tomography'. Together they form a unique fingerprint.

Cite this