A model based algorithm for perfusion estimation in interventional C-arm CT systems

• Verfasst von: Wagner, Martin [VerfasserIn]
• Verfasst von: Deuerling-Zheng, Yu [VerfasserIn]
• Verfasst von: Möhlenbruch, Markus Alfred [VerfasserIn]
• Verfasst von: Bendszus, Martin [VerfasserIn]
• Verfasst von: Boese, Jan [VerfasserIn]
• Verfasst von: Heiland, Sabine [VerfasserIn]
• Titel: A model based algorithm for perfusion estimation in interventional C-arm CT systems
• Verf.angabe: Martin Wagner, Yu Deuerling-Zheng, Markus Möhlenbruch, Martin Bendszus, Jan Boese, Sabine Heiland
• E-Jahr: 2013
• Jahr: 1 March 2013
• Umfang: 11 S.
• Teil: volume:40
• Teil: year:2013
• Teil: number:3
• Teil: elocationid:031916
• Teil: pages:1-11
• Teil: extent:11
• Fussnoten: Gesehen am 23.09.2021
• Titel Quelle: Enthalten in: Medical physics
• Ort Quelle: Hoboken, NJ : Wiley, 1974
• Jahr Quelle: 2013
• Band/Heft Quelle: 40(2013), 3, Artikel-ID 031916, Seite 1-11
• ISSN Quelle: 2473-4209
• ISSN Quelle: 1522-8541
• DOI: doi:10.1118/1.4790467
• Datenträger: Online-Ressource
• Sprache: eng
• Sach-SW: backpropagation
• Sach-SW: Computed tomography
• Sach-SW: Computerised tomographs
• Sach-SW: computerised tomography
• Sach-SW: computerized tomography
• Sach-SW: Digital computing or data processing equipment or methods
• Sach-SW: dynamic reconstruction
• Sach-SW: Flow visualization
• Sach-SW: Fluid transport and rheology
• Sach-SW: Haemodynamics
• Sach-SW: haemorheology
• Sach-SW: Image data processing or generation
• Sach-SW: image reconstruction
• Sach-SW: Image reconstruction
• Sach-SW: image segmentation
• Sach-SW: in general
• Sach-SW: Interpolation
• Sach-SW: iterative methods
• Sach-SW: iterative reconstruction
• Sach-SW: Medical image noise
• Sach-SW: medical image processing
• Sach-SW: Medical image quality
• Sach-SW: Medical image reconstruction
• Sach-SW: Medical imaging
• Sach-SW: model function
• Sach-SW: perfusion imaging
• Sach-SW: Reconstruction
• Sach-SW: Segmentation
• Sach-SW: specially adapted for specific applications
• Sach-SW: Tissues
• K10plus-PPN: 177169937X

Abstract

Purpose: Interventional C-arm CT imaging, today, plays an important role in the diagnosis and treatment of patients. The main part of the 3D imaging techniques, currently used in interventions, are morphological imaging techniques. So far, the ability for functional or perfusion imaging is limited, e.g., only static cerebral blood volume measurement [A. S. Ahmed, Y. Deuerling-Zheng, C. M. Strother, K. A. Pulfer, M. Zellerhoff, T. Redel, K. Royalty, D. Consigny, M. J. Lindstrom, and D. B. Niemann, “Impact of intra-arterial injection parameters on arterial, capillary, and venous time-concentration curves in a ca480 nine model,” AJNR Am. J. Neuroradiol. 30, - (2009) 10.3174/ajnr.A1586] is available. The sample rate of current C-arm CT systems is not fast enough yet to measure dynamic parameters like cerebral blood flow using standard Feldkamp reconstruction. Methods: The authors propose a reconstruction algorithm that models the time-dependent attenuation values of each voxel using a gamma-variate function. The method can be divided into a segmentation-based initialization and an iterative optimization step. For the initialization, a threshold-based segmentation of vessel, tissue, and nondynamic structures (e.g., bone and air) is performed on the filtered backprojection (FBP) reconstructions. For each of these regions, homogeneous time-attenuation curves are estimated to initialize all the voxels within the region. The scaling-factor is then adjusted for each voxel using the attenuation values of the static reconstructions. The second part of the algorithm is an iterative optimization of the gamma-variate parameters of each voxel, based on a simultaneous algebraic reconstruction technique. Within each iteration, a Levenberg optimization is applied to minimize the backprojected errors. Results: The algorithm is quantitatively evaluated with simulated forward projections as well as real C-arm CT projection data. In the phantom experiments, penumbra and infarct core could be segmented with an adjusted Rand index of up to 0.95 for a noise level of 105 photons. Perfusion CT data sets from three patients were used to compare the iterative reconstruction approach to the interpolated FBP reconstruction using different sweep times. In their experiments, a sweep time of 4 s using iterative reconstruction would be equivalent to that using interpolated FBP with a sweep time of around 1 s. The reconstruction results of the animal study are compared to a perfusion CT acquisition, sampled with 1 frame per second. A correlation coefficient of 0.75 between the original and the reconstructed CBF-maps could be reached with the iterative approach compared to 0.56 using the interpolated FBP reconstruction. Conclusions: In their experiments, the quality of dynamic perfusion measurements was improved using the proposed reconstruction algorithm compared to static reconstruction followed by interpolation. It could be used to increase the temporal resolution of current C-arm CT system without hardware modification to make them feasible for dynamic perfusion measurement. Furthermore, radiation dose could be reduced using their method to increase temporal resolution than using static reconstruction with a higher sampling frequency.