Software architecture for multi-bed FDK-based reconstruction in X-ray CT scanners

Abstract

Cone-beam t r a c tMost small-animal X-ray computed tomography (CT) scanners are based on cone-beam geometry with a flat-panel detector orbiting in a circular trajectory. Image reconstruction in these systems is usually performed by approximate methods based on the algorithm proposed by Feldkamp et al. (FDK). Besides the implementation of the reconstruction algorithm itself, in order to design a real system it is necessary to take into account numerous issues so as to obtain the best quality images from the acquired data. This work presents a comprehensive, novel software architecture for small-animal CT scanners based on cone-beam geometry with circular scanning trajectory. The proposed architecture covers all the steps from the system calibration to the volume reconstruction and conversion into Hounsfield units. It includes an efficient implementation of an FDK-based reconstruction algorithm that takes advantage of system symmetries and allows for parallel reconstruction using a multiprocessor computer. Strategies for calibration and artifact correction are discussed to justify the strategies adopted. New procedures for multi-bed misalignment, beam-hardening, and Housfield units calibration are proposed. Experiments with phantoms and real data showed the suitability of the proposed software architecture for an X-ray small animal CT based on cone-beam geometry. IntroductionMany small animal X-ray computed tomography (CT) scanners are based on cone-beam geometry with a flat-panel detector orbiting in a circular trajectory [1][2][3][4]. This configuration presents advantages over other alternatives used in clinical and preclinical applications: reduction of acquisition time, large axial field of view (FOV) without geometrical distortions, and optimization of radiated dose [5]. proposed by Feldkamp et al. (FDK) [6] are still widely used for solving the 3D reconstruction task because of their straightforward implementation and computational efficiency [4]. Almost every aspect of the reconstruction process has been studied: there is literature on algorithm variations for different trajectories [7,8], optimizations using graphic processing units (GPUs) [9][10][11][12][13][14][15], strategies to reduce cone beam artifacts [16,17], study of consistency conditions [18], optimization of the back-projection step [19], etc. However, in a real practical system, the implementation of a reconstruction algorithm core such as FDK is just an initial step of the process, and there 1

Functional segmentation of dynamic PET studies: Open source implementation and validation of a leader-follower-based algorithm

“…The metric of choice is generally the Euclidean distance; however, since this distance may yield quite different values, it is more convenient for this particular algorithm to use a metric for which a known threshold can be set. For instance, it is possible to use the Pearson or Spearman correlation coefficients and the cosine product, which have values bound in [ 1,1] and measure the similarity in shape between two given vectors. For this kind of metric, a voxel is included in a given cluster if the similarity measurement is above the threshold.…”

“…This similarity metric computes the cosine of the angle between two given vectors. Its values are bound in [ 1,1], which allows the user to set a sensible and meaningful threshold. where D(v, c i ) is the Euclidean distance between the voxel TAC vector v and the centroid from cluster i.…”

“…CT images were reconstructed using a Feldkamp algorithm to obtain an isotropic voxel size of 0.125 mm [1]. These anatomical images did not require any registration with their corresponding PET scans thanks to the intrinsic alignment of the PET/CT device.…”

A method for small-animal PET/CT alignment calibration

“…Each CT scan was reconstructed using a specific FDK-based software (Abella et al 2011) into CT datasets with matrix size 256 × 256 × 256 mm 3 and voxel size 0.123 × 0.123 × 0.123 mm 3 . PET scan was reconstructed using the 2D-OSEM algorithm available in the Argus scanner into 175 × 175 × 61 mm 3 matrix size datasets with a voxel size of 0.3875 × 0.3875 × 0.775 mm 3 .…”

Investigation of different Compressed Sensing approaches for respiratory gating in small animal CT

“…Low-dose data were reconstructed by using Mangoose [11], an approximation of the FDK method, and three algorithms based on the Split-Bregman approach: TV-PICCS, Ll-PICCS, WT -PICCS. Parameters in each algorithm were selected to give comparable resolution properties.…”