Jeffrey Brokish

Sampling Requirements for Circular Cone Beam Tomography

The sampling requirements of tomographic projection data are determined by the set of frequencies occupied by the Fourier transform of the projection data. This region of essential support has been analyzed for various 2-D geometries. The general case of 3-D cone beam projections, using a 2-D detector array, is mainly unexplored. In this paper, we consider the 3-D circular cone-beam (CCB) scan. An analysis of the essential support of CCB projection data provides the sampling requirements for detector spacing, row spacing, and projection count (views).

A hierarchical algorithm for fast backprojection in helical cone-beam tomography

Existing algorithms for exact helical cone beam (HCB) tomographic reconstruction involve a 3-D backprojection step, which dominates the computational cost of the algorithm. We present a fast hierarchical 3-D backprojection algorithm, generalizing fast 2-D parallel beam and fan beam algorithms, which reduces the complexity of this step from O(N/sup 4/) to O(N/sup 3/log N), greatly accelerating the reconstruction process.

Evolving roads in IKONOS multispectral imagery

The introduction of satellite imagery characterized by high spectral and spatial resolutions has made possible the development of new viable approaches for the accurate, and cost-effective extraction of linear features with minimal human intervention. This paper presents a semi-automated method for the extraction of roads from (1-meter) pan-sharpened multispectral IKONOS imagery. An operator provides an initial seed point on the road of interest, then the region is evolved using a level set method. Further analysis through iterative smoothing refines the extracted region to accurately estimate the road centerline despite the presence of cars on the road, changes in the pavement or surface properties of the road, or obstruction resulting from foliage or shadows cast on the road by neighboring trees. Initial results have demonstrated the utility of the algorithm in efficiently extracting roads from high resolution satellite imagery with minimal human interaction. Over 97% delineation accuracy was achieved on manually ground truthed IKONOS image samples overlooking both urban and rural locations.

Ultra-fast hierarchical backprojection for Micro-CT reconstruction

This paper introduces a new version of a fast and accurate O(N3logN) reconstruction algorithm for cone-beam CT with a circular source trajectory. This is the first report on the performance of this algorithm with high-resolution large-matrix Micro-CT data. We determined that a software implementation of the FHBP algorithm on off-the-shelf workstations can provide speedups by an order of magnitude or more in Micro-CT reconstruction with no perceptible loss of image quality as compared to state-of-the-art commercial software implementations of conventional reconstruction. This is especially important for Micro-CT reconstructions, where maximum image size can be as large as 8 Ktimes8 K pixels per slice and reconstruction of such image volumes can span days. Owing to its algorithmic acceleration, the FHBP-based code achieves the performance of a cluster of computers on a single PC, and could provide even faster reconstruction rates if running on the entire cluster.

Fast hierarchical backprojection for helical cone-beam tomography

Existing algorithms for exact helical cone beam (HCB) tomographic reconstruction are computationally infeasible for clinical applications. Their computational cost is dominated by 3-D backprojection, which is generally an O(N/sup 4/) operation. We present a fast hierarchical 3-D backprojection algorithm, generalizing fast 2-D parallel beam and fan beam algorithms, which reduces the overall complexity of this step to O(N/sup 3/ log N), greatly accelerating the reconstruction.

Combined algorithmic and hardware acceleration for ultra-fast backprojection

We describe the first implementation and performance of a fast O(N2logN) hierarchical backprojection (FHBP) algorithm on a field programmable gate array (FPGA) platform. The resulting prototype tomographic backprojection system for 2D fan-beam geometry combines speedup through algorithmic improvements provided by the FHBP algorithm, with speedup of a specialized hardware platform. For data parameters typical in diagnostic CT, and using an off-the-shelf FPGA evaluation board, we report normalized reconstruction speeds of 160 frames a second, and relative speedup of 25x compared to conventional backprojection on the same hardware. These results demonstrate that the reduction in operation counts demonstrated previously for the FHBP algorithm can be translated to a comparable actual run time improvement in a massively parallel hardware implementation, while preserving stringent diagnostic image quality. The dramatic speedup and throughput achieved indicate the feasibility of systems based on this technology, which achieve real-time 3D reconstruction for state-of-the art diagnostic CT scanners at low power, small footprint, high-reliability, and affordable cost.

Iterative circular conebeam CT reconstruction using fast hierarchical backprojection/reprojection operators

This is the first report on a new fast statistical iterative reconstruction algorithm for conebeam with a circular source trajectory, accelerated by InstaRecons fast O(N3logN) hierarchical cone beam backprojection1 and reprojection algorithms. We report on the results of image quality and run-time comparisons with iterative algorithms based on conventional backprojection and reprojection. We demonstrate that the iterative algorithm introduced here can provide image quality indistinguishable from an iterative algorithm using conventional BP/RP operators, while providing almost a 10x speedup in reconstruction rates. Combining the 10x algorithmic acceleration with additional hardware acceleration by FPGA, Cell, or GPU implementation, this work indicates the feasibility of iterative reconstruction algorithms for dose reduction and image quality improvement in routine CT practice, at competitive speeds and affordable cost.

Combined algorithmic and GPU acceleration for ultra-fast circular conebeam backprojection

In this paper, we describe the first implementation and performance of a fast O(N3logN) hierarchical backprojection algorithm for cone beam CT with a circular trajectory1,developed on a modern Graphics Processing Unit (GPU). The resulting tomographic backprojection system for 3D cone beam geometry combines speedup through algorithmic improvements provided by the hierarchical backprojection algorithm with speedup from a massively parallel hardware accelerator. For data parameters typical in diagnostic CT and using a mid-range GPU card, we report reconstruction speeds of up to 360 frames per second, and relative speedup of almost 6x compared to conventional backprojection on the same hardware. The significance of these results is twofold. First, they demonstrate that the reduction in operation counts demonstrated previously for the FHBP algorithm can be translated to a comparable run-time improvement in a massively parallel hardware implementation, while preserving stringent diagnostic image quality. Second, the dramatic speedup and throughput numbers achieved indicate the feasibility of systems based on this technology, which achieve real-time 3D reconstruction for state-of-the art diagnostic CT scanners with small footprint, high-reliability, and affordable cost.

Noise performance of fast hierarchical 3D backprojection for helical cone-beam tomography

Existing algorithms for exact helical cone beam tomographic reconstruction involve a 3-D backprojection step, which dominates the computational cost of the algorithm. Hierarchical backprojection reduces the complexity of this step from O(N/sup 4/) to O(N/sup 3/logN), greatly accelerating the reconstruction process. Here the performance of the hierarchical reconstruction is examined in the presence of noise. We demonstrate that reconstructions obtained using this method have good image quality and comparable noise performance to conventional backprojection, while providing a speedup in computation by over an order of magnitude. These properties are essential for acceptance of a fast reconstruction algorithm.