Search

We developed a method to evaluate the accuracy of segmentation algorithms. Oversegmentation, undersegmentation, missing and spurious labels may all appear concurrently in machine segmented images. Segmentation algorithms make systematic errors and have different optimal operating ranges. Existing methods of segmentation evaluation do not evaluate these details. Our method, based on multiple feature recovery, reports systematic errors and indicates optimal operating ranges of features, besides measuring overall errors. A knowledge of the magnitude and type of errors can be used for tuning or selecting segmentation algorithms. Although our method was developed for CT scanning for security, it is applicable to other fields, including medical imaging, where multi-object feature recovery, non-uniform costs and a knowledge of optimal operating ranges are helpful.

Author(s): Karimi, Seemeen; Jiang, Xiaoqian; Cosman, Pamela; Martz, Harry | Abstract: BackgroundImaging systems used in aviation security include segmentation algorithms in an automatic threat recognition pipeline. The segmentation algorithms evolve in response to emerging threats and changing performance requirements. Analysis of segmentation algorithms' behavior, including the nature of errors and feature recovery, facilitates their development. However, evaluation methods from the literature provide limited characterization of the segmentation algorithms.ObjectiveTo develop segmentation evaluation methods that measure systematic errors such as oversegmentation and undersegmentation, outliers, and overall errors. The methods must measure feature recovery and allow us to prioritize segments.MethodsWe developed two complementary evaluation methods using statistical techniques and information theory. We also created a semi-automatic method to define ground truth from 3D images. We applied our methods to evaluate five segmentation algorithms developed for CT luggage screening. We validated our methods with synthetic problems and an observer evaluation.ResultsBoth methods selected the same best segmentation algorithm. Human evaluation confirmed the findings. The measurement of systematic errors and prioritization helped in understanding the behavior of each segmentation algorithm.ConclusionsOur evaluation methods allow us to measure and explain the accuracy of segmentation algorithms.

Metal artifact reduction methods in computed tomography replace the projection data passing through metals with an estimate of the true data. Inaccurate estimation leads to the generation of secondary artifacts. Data estimates can be improved by the use of prior knowledge of the projection data. In this paper, a method has been created to generate a prior image. The method uses computer vision techniques to segment regions of the initially reconstructed image and then discriminates between regions that are likely to be artifacts and anatomical structures. Results on test images show that metal artifacts are reduced and that few secondary artifacts are present, even in the case of multiple metal objects.

We are developing a new technique using both active (A) and passive (P) computed tomography (CT) to overcome problems in the quantitative analysis of plutonium contents in the molten salt extract (MSE) plutonium buttons. Our first three dimensional image of the MSE button using ACT scan has already shown heterogeneous behavior throughout the button as opposed to the model of a homogeneous button as previously thought. An image from a PCT scan, although it does not have sufficient spatial resolution and data, also seems to support the heterogeneous picture.

The US Department of Homeland Security (DHS) has requirements for future explosives scanners that include dealing with a larger number of threats, higher probability of detection, lower false alarm rates and lower operating costs. One tactic that DHS is pursuing to achieve these requirements is to augment the capabilities of the established security vendors with third-party algorithm developers. The purposes of this presentation are to review DHS's objectives for involving third parties in the development of advanced algorithms and then to discuss how these objectives are achieved using workshops and grand challenges.

The purpose of this Standard Operating Procedure (SOP) is to give guidelines on how to determine the density of a sample that will be used as the requirement density. This will be the requirement density of record for the specimens examined by Micro CT and EDS measurements. This density will then be set as the formulation requirement for radiography measurements. This SOP is referred to in TP 48— Preparation of Hydrogen Peroxide/Icing Sugar Specimens for X-ray Measurements by J. G. Reynolds and H. E. Martz.

The purpose of this Standard Operating Procedure (SOP) is to give guidelines on how to measure the volume of a container for use in radiography studies. This SOP gives details on a recommended method of how exactly to measure the volume of containers used for specimens for Micro CT and EDS measurements. This volume is important in determining accurate densities of the specimens. This SOP is referred to in TP 48 - Preparation of Hydrogen Peroxide/Icing Sugar Specimens for X-ray Measurements by J.G. Reynolds and H.E. Martz

Gamma-ray computed tomography (CT) is potentially powerful nondestructive method for assessing the degree of distress that exists in reinforced-concrete structures. In a study to determine the feasibility of using CT to inspect reinforced-concrete specimens, we verified that CT can quantitatively image the internal details of reinforced concrete. To assess the accuracy of CT in determining voids and cracks, we inspected two fiber-reinforced concrete cylinders (one loaded and one unloaded) and a third cylinder containing a single reinforcing bar (rebar). To evaluate the accuracy of CT in establishing the location of reinforcing bars, we also inspected a concrete block containing rebars with different diameters. The results indicate that CT was able to revolve the many different phases in reinforced concrete (voids, cracks, rebars, and concrete) with great accuracy. 15 refs., 10 figs.

X-ray computerized tomography (CAT or CT) is a sophisticated imaging technique that provides cross-sectional views of materials, components and assemblies for industrial nondestructive evaluation (NDE). We have studied the feasibility of using CT as an inspection tool for reinforced concrete and the use of multi-energy, linear, attenuation techniques to deduce variations in density (ρ) and/or atomic number (Z) that could be caused by varying the types of concrete mixes and/or compaction in the concrete itself. To perform this study, we designed and built a prototype medium-/high-energy (200- to 2000 keV) CT scanner — ZCAT — to image small concrete samples (± 30 cm in diameter and ± 75 cm in height) with a spatial resolution of about 2 mm. We used ZCAT to quantitatively inspect a 20 cm concrete cube with 1.27 cm diameter reinforcing bars (rebars) and to measure p and/or Z variations in a 20 cm diameter concrete cylinder. We describe the ZCAT scanner design, some of its physical limitations and the data-acquisition parameters used in our study. Our results and those of others [1,2] show that CT can be used to inspect reinforced concrete and to distinguish material p and/or Z variations within concrete.

An efficient method for accurately calculating the center-of-rotation, or projection center, for parallel computed tomography projection data, or sinograms, is described. This method uses all the data in the sinogram to estimate the center by a least-squares technique and requires no previous calibration scan. The method also finds the object's center-of-mass without reconstructing its image. Since the method uses the measured data, it is sensitive to noise in the measurements, but that sensitivity is relatively small compared to other techniques. Examples of its use on simulated and actual data are included. For fan-beam data over 360 degrees , two related methods are described to find the center in the presence or absence of a midline offset. >

x-Ray computed axial tomography (CT) provides cross-sectional views of materials, components, and assemblies for industrial non-destructive evaluation. We have applied CT imaging to quantitatively measure the 3-D distribution ogf x-ray attenuation at reasonably high resolutions. In our industrial x-ray CT-studies, we have centered on two technical approaches: a first-generation translate/rotate CT system that consist of well-collimated (∼ 0.55 mm) photon source detector, and a third-generation scanner that uses a fluoroscopy detector.

The mix between x-ray phase and attenuation information needs to be understood for accurate object recovery from radiography and tomography data. We are researching and experimentally validating algorithms that simulate x-ray phase contrast to determine the required physics necessary for quantitative object recovery. The results of a study are described to determine if a multislice (beam-propagation) method is required for simulating x-ray radiographs. We conclude that the multislice method is not required for accurate simulation of greater than or equal to 8 keV x-ray radiographs of millimeter-size objects with micrometer structures.

Summary form only given. We demonstrate phase-contrast enhanced X-ray imaging (PCXI) for characterization of the solid deuterium-tritium surface inside of beryllium shells. This novel method of X-ray imaging takes advantage of the small refraction and diffraction of the X-rays at boundaries between materials. PCXI is more sensitive to small density changes in low atomic number materials than traditional absorption radiography. Point-projection X-ray microscopes were used in our laboratory to study the surface roughness of a solid deuterium-tritium layer inside of a beryllium capsule. This method can also be applied to plasma physics characterization

High energy density experiments, such as those planned at the National Ignition Facility (NIF), use mesoscale targets with the goals of studying high energy density physics, inertial confinement fusion, and the support of national security needs. Mesoscale targets are typically several millimeters in size and have complex micrometer‐sized structures composed of high‐density metals and low‐density foams and ices. These targets are designed with exacting tolerances that are difficult to achieve at present. Deviation from these tolerances can result in compromise of experimental goals and thus it is necessary to determine as‐built properties of these targets using NDE techniques. Radiography and computed tomography are being used to investigate these targets, but the mix between phase and absorption information is difficult to separate, making interpretation of results difficult. We have recently improved the HADES radiographic simulation code to include phase in simulations, as an aid for doing NDE on mesos...

A high-speed Radon transform computer (RTC) has been designed and simulated that operates on the principles of the parallel projection pipeline engine (PPPE) processor suggested by J.L.C. Sanz, et al. (1988). The computer consists of an array of identical processing elements with fast communication links to each of its neighbors and uses forward- and back-projection formulas tests linear interpolation. Simulation and experimental timing tests have been performed on both a systolic array of 64 transputer processors and on arrays of commercial DSP (digital signal processor) chips. A custom VLSI implementation is currently under design which is expected to perform forward and inverse transforms of 512 by 512 images at or near video rates. >

A class of hybrid domain-iterative algorithms for 2-D image reconstruction is described. The consistent iterative reconstruction-reprojection (CIRR) algorithm was designed and implemented along with other classical domain-iterative methods. The algorithms are tested on the hollow projections problem, where simple object models are assumed known. Results show that the CIRR algorithm is robust and accurate, and converges quickly. >

The failure of a steel pipe subjected to shock loading was observed using x ray imaging. We describe and analyze the x ray images in detail. We see radiographic evidence that most of the fractures were due to shear rather than brittle failure. We also make quantitative comparisons between static radiographs and simulations but do not see perfect agreement. The sources of the current lack of agreement are discussed, as well as future work planned.

We have developed a new tomography code, HADES-CCG. This code uses HADES, a radiographic simulation code, to perform forward- and back-projection and is coupled to a Constrained Conjugate Gradient (CCG) optimizer. An iterative solution to the reconstruction problem is found which is optimal, given the detector noise model, a source model and the appropriate attenuation cross-sections. By explicitly including experimental effects in forward- and back-projection, these effects are not folded back into the object model.

The development and improvement of advanced materials is strictly connected to the understanding of the properties and behavior of such materials as a function of both their macro and micro-structures. The application of X-ray computed tomography (CT) to these materials allows for a better understanding of the materials properties and behavior on either macro or micro-structure scales. The authors applied CT to study a set of aerospace grade carbon fiber/thermoplastic matrix composites. Samples of APC-2 (PEEK/AS4) were subjected to either static or high-stress fatigue loading in tension. Both notched (central circular hole) and unnotched specimens were examined. They are investigating a high-temperature thermoplastic polyimide composite sample by acquiring CT data sets before, during (at set intervals), and after full-reversal (tension-compression), low-stress fatigue loading at the upper use temperature. The CT scanner employed and the results obtained in the analysis of 3D CT data sets to study the defects and other features within the different composites are presented in this report.

Computed Tomography (CT) using penetrating radiation (x- or gamma-rays) can be used in a number of aircraft applications. This technique results in 3D volumetric attenuation data that is related to density and effective atomic number. CT is a transmission scanning method that must allow complete access to both sides of the object under inspection; the radiation source and detection systems must surround the object. This normally precludes the inspection of some large or planar (large aspect ratio) parts of the aircraft. However, we are pursuing recent limited-data techniques using object model information to obtain useful data from the partial information acquired. As illustrative examples, we describe how CT was instrumental in the analysis of particular aircraft components. These include fuselage panels, single crystal turbine blades, and aluminum-lithium composites.© (1993) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Members of the Nondestructive Evaluation Section at the Lawrence Livermore National Laboratory (LLNL) are implementing the advanced three-dimensional imaging technique of x- and gamma-ray computed tomography (CAT or CT) for industrial and scientific obscured object evaluation. This technique provides internal and external views of materials, components, and assemblies nonintrusively. Our work includes building of CT scanners as well as data preprocessing, image reconstruction, display and analysis algorithms. These capabilities have been applied to a variety of industrial and scientific NDE applications. We have used CT to study various objects with obscured features at our laboratory ranging in size from 1 mm3 to 1 m3. In these studies, CT has revealed flaws (e.g. cracks, voids, and inclusions), internal and external dimensional information, differences in elemental composition or material density, and other important material characteristics. CT has also been used to localize, identify, and quantify radioisotopes within canisters. As illustrative examples, we describe how CT was instrumental in the analysis of concrete specimens, diesel engine thermocouple plugs, jet engine turbine blades, ballistic target materials, and radioactive waste canisters.© (1993) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Considerable effort is being applied to develop multiple-energy industrial CT techniques for materials characterization. Multiple-energy CT can provide reliable estimates of effective Z (Z{sub eff}), weight fraction, and rigorous calculations of absolute density, all at the spatial resolution of the scanner. Currently, a wide variety of techniques exist for CT scanners, but each has certain problems and limitations. Ultimately, the best multi-energy CT technique would combine the qualities of accuracy, reliability, and wide range of application, and would require the smallest number of additional measurements. We have developed techniques for calculating material properties of industrial objects that differ somewhat from currently used methods. In this paper, we present our methods for calculating Z{sub eff}, weight fraction, and density. We begin with the simplest case -- methods for multiple-energy CT using isotopic sources -- and proceed to multiple-energy work with x-ray machine sources. The methods discussed here are illustrated on CT scans of PBX-9502 high explosives, a lexan-aluminum phantom, and a cylinder of glass beads used in a preliminary study to determine if CT can resolve three phases: air, water, and a high-Z oil. In the CT project at LLNL, we have constructed several CT scanners of varying scanning geometries usingmore » {gamma}- and x-ray sources. In our research, we employed two of these scanners: pencil-beam CAT for CT data using isotopic sources and video-CAT equipped with an IRT micro-focal x-ray machine source.« less

Proton beams with energies of 5 and 7 MeV are focused to 5 μm and used to produce tomograms of capillary tubes and low-density foams. In this energy range, proton energy loss is primarily due to interactions with electrons. Therefore, by measuring the residual energy of protons transmitted through samples in a manner similar to that used for Scanning Transmission Ion Microscopy (STIM), and reconstructing a cross-sectional image from multiple projections, we can map out spatial variations in electron density due to sample geometry and composition. In our experimental arrangement, the sample is translated and rotated in a stationary proton beam. Transmitted proton energies are measured using a silicon surface barrier detector. Tomographie reconstructions are produced from the calculated line-average densities using a procedure based on a filtered backprojection algorithm developed for X-ray computed tomography (CT) systems. The technique is especially useful in characterizing samples where large variations in Z or low total density limit the applicability of X-ray CT analysis.