Imaging properties of a rotation-free, arrayed-source micro-computed tomography system

medical imaging;x-ray;image reconstruction;computed tomography

Quan, Enzhuo ; David Lalush, Committee Chair,Caterina Gallippi, Committee Member,Mohamed Bourham, Committee Member,Jianping Lu, Committee Member,Wesley Snyder, Committee Member,Quan, Enzhuo ; David Lalush ; Committee Chair ; Caterina Gallippi ; Committee Member ; Mohamed Bourham ; Committee Member ; Jianping Lu ; Committee Member ; Wesley Snyder ; Committee Member

University:North Carolina State University

关键词: medical imaging;    x-ray;    image reconstruction;    computed tomography;   

Others  :  https://repository.lib.ncsu.edu/bitstream/handle/1840.16/3325/etd.pdf?sequence=1&isAllowed=y

美国|英语

来源: null

【 摘 要 】

We study the three-dimensional reconstructions and imaging properties of a proposedrotation-free micro-computed tomography (CT) system. The system uses linear arrays of thecarbon nano-tube (CNT)-based X-ray sources which have ultra-short switch time and areindividually addressable. With such sources, the micro-CT system is able to achieve ultrahightemporal resolution, reduce dose and facilitate gated imaging. A square and a hexagonalgeometry have been proposed for the system. In the square geometry, two linear sourcearrays and two area detectors form a square; whereas in the hexagonal geometry, three linearsource arrays and three area detectors form a hexagon. The tomographic angular sampling forboth of the geometries requires no motion of the sources or subject. Based on the sinogrammaps, the hexagonal geometry has improved angular coverage than the square geometry. Theordered-subset convex iterative algorithm is implemented in both geometries forreconstructions from cone-beam projection data. Mean squared errors at the uniform regionsin the reconstructed images are calculated to quantify the artifact level. Point spreadfunctions are examined for point objects located at different axial and transverse positionsthroughout the FOV. Variance images are generated from 100 reconstructions with simulatedPoisson noise and the mean variance are calculated for different regions of interest. Theeffect of gaps between the source arrays and the detectors is also studied. The reconstructedimages from both geometries are generally consistent with the phantom, although somestreaking artifacts due to the limited-angle nature of the geometries are observed. The gapfreehexagonal geometry produces lower mean squared error in the reconstructed images,lower FWHM of the point spread functions and lower variance. However, in more realisticsituations where gaps appear between the source arrays and the detectors, the angularcoverage of the hexagonal geometry degrades faster, resulting in an increase in artifacts, sothat the square geometry becomes superior in this case.