High resolution iterative CT reconstruction using graphics hardware

Abstract

Abstract-Particular applications of computed tomography require high slice resolutions. The fastest iterative implementations on graphics cards use 3-D textures to exploit hardwareaccelerated trilinear interpolation. However, the size of 3-D textures is subject to technical limitations, which makes them inapplicable here. Alternatively a 2-D texture array can be used instead of the 3-D texture as in early graphics implementations. The additional memory synchronizations cause a significant loss of performance.We utilize new features of the recently released CUDA 2.2 framework to improve the performance of the Simultaneous Algebraic Reconstruction Technique (SART). In this paper we present an enhanced version of our efficient implementation of the most time-consuming parts of the iterative reconstruction algorithm: forward-and back-projection.We explain the required strategy to adapt the algorithm for the CUDA 2.2 features, in particular the usage of 2-D texture lookups from pitchlinear memory. Finally, we compare the result to our previous ones with respect to both reconstruction speed and technical limitations. The proposed strategy is a new balance between performance and limitations in resolution.

Development of a Dedicated Radiotherapy Unit with Real-Time Image Guidance and Motion Management for Accelerated Partial Breast Irradiation

“…It is well known that the full-fan, half-scan CBCT using the FDK algorithm suffers more severe cone-beam artifacts than the full-fan, full-scan CBCT for slices away from the central slice due to increased missing data. 29 This phenomenon (reduced images quality for the half-scan CBCT), however, was not observed in this study. As shown in Fig.…”

“…cone-beam artifacts), particularly for half-scan acquisition because the cone beam projection images acquired in a circular trajectory do not completely cover the Fourier space and do not provide complete data. 28,29 Another popular method for CT/CBCT reconstruction is the simultaneous algebraic reconstruction technique (SART)an algebraic reconstruction method solving the linear system using iterative methods without direct matrix inversion. In comparison to the FBP approach, the algebraic method is generally more advantageous in CT and CBCT reconstruction using incomplete data because the algebraic method is easy to implement for different scanning geometries.…”

“…The CUDA-based method for accelerating image reconstruction was used in many prior studies [24][25][26][27][28][29][30][31][32]. It is still an open question whether the CUDA-based approach outperforms the graphicsbased approach.…”

A fast forward projection using multithreads for multirays on GPUs in medical image reconstruction

“…In their algorithms, one thread was assigned to one rod of projection and one rod of volume, respectively. Keck et al 29 presented an enhanced version of efficient implementation for the forward-and backprojection of iterative reconstruction algorithms using NVIDIA QuadroFX 5600 and Tesla C1060. Different from our work, they adapted the usage of two-dimensional texture lookups from pitchlinear memory to overcome the threedimensional texture size limitation for volume representation.…”

Fast polyenergetic forward projection for image formation using OpenCL on a heterogeneous parallel computing platform

“…The CUDAbased method for accelerating image reconstruction was used in many prior studies. [24][25][26][27][28][29][30][31][32] It is still an open question whether the CUDA-based approach outperforms the graphics-based approach.…”