Abstract:Abstract Objective: To introduce the optimization of a customized GPU-based simultaneous algebraic reconstruction technique (cSART) in the field of phase-contrast breast computed tomography (bCT). The presented algorithm features a 3D bilateral regularization filter that can be tuned to yield optimal performance for clinical image visualization and tissues segmentation. Approach: Acquisitions of a dedicated test object and a breast specimen were performed at Elettra, the Italian synchrotron radiation (SR) facility (Trieste, Italy) using a large area CdTe single-photon counting detector. Tomographic images where obtained at 5 mGy of mean glandular dose, with a 32 keV monochromatic X-ray beam in the free-space propagation mode. Three independent algorithm’s parameters were optimized by using contrast-to-noise ratio (CNR), spatial resolution, and noise texture metrics. The results obtained with the cSART algorithm were compared with conventional SART and filtered back projection (FBP) reconstructions. Image segmentation was performed both with gray scale-based and supervised machine-learning approaches. Main results: Compared to conventional FBP reconstructions, results indicate that the proposed algorithm can yield images with a higher CNR (by 35% or more), retaining a high spatial resolution while preserving their textural properties. Alternatively, at the cost of an increased image “patchiness”, the cSART can be tuned to achieve a high-quality tissue segmentation, suggesting the possibility of performing an accurate glandularity estimation potentially of use in the realization of realistic 3D breast models starting from low radiation dose images. Significance: The study indicates that dedicated iterative reconstruction techniques could provide significant advantages in phase-contrast bCT imaging. The proposed algorithm offers great flexibility in terms of image reconstruction optimization, either towards diagnostic evaluation or towards image segmentation.

SART Algorithm Parallel Implementation with Siddon and MBR Methods for Digital Linear Tomosynthesis Using OpenCL

Fast Projection Algorithm for LIM-Based Simultaneous Algebraic Reconstruction Technique and Its Parallel Implementation on GPU.

CUDA Based Parallel Implementation of Simultaneous Algebraic Reconstruction Technique

PARALLELPROJ -- An open-source framework for fast calculation of projections in tomography

Fast parallel implementation for total variation constrained algebraic reconstruction technique

Feasibility of GPU-assisted Iterative Image Reconstruction for Mobile C-arm CT

A Parallel Reconstruction Algorithm Based on Curvilinear Projection Model in Tomography

Parallel Multi-threaded Gridrec Algorithm for Computer Tomography on GPU for Edge Computing

Parallel Implementation of Katsevich's FBP Algorithm

Parallel SART Algorithm of Linear Scan Cone-Beam CT for Fixed Pipeline

Accelerated Cone Beam CT Reconstruction Based on OpenCL

Toward Parallel Optimal Computation of Ultrasound Computed Tomography Using GPU

A Parallel Implementation of the Katsevich Algorithm for 3-D CT Image Reconstruction

Modified Simultaneous Algebraic Reconstruction Technique and its Parallelization in Cryo-electron Tomography

Optimization of a customized simultaneous algebraic reconstruction technique algorithm for phase-contrast breast computed tomography

Accelerated CT Reconstruction Using GPU SIMD Parallel Computing with Bilinear Warping Method

An Effective Cuda Parallelization Of Projection In Iterative Tomography Reconstruction

Iterative Ct Reconstruction Integrating Sart and Conjugate Gradient

Variable step size methods for solving simultaneous algebraic reconstruction technique (SART)-type cbct reconstructions

Convergence of iterative image reconstruction algorithms for Digital Breast Tomosynthesis

Fast Katsevich algorithm based on GPU for helical cone-beam computed tomography.