Search

Deep learning image reconstruction (DLIR) is a most recent computed tomography (CT) image reconstruction method for reducing image noise while maintaining the spatial resolution of the image. It is well-understood that image noise affects the low-contrast detectability (LCD) of the images. We assessed the LCD of DLIR images using parameters of CT number accuracy, signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR) and visual assessment on the low-contrast module in a Catphan phantom and compared it to images reconstructed with filtered-back projection (FBP). We scanned the Catphan 604 phantom (The Phantom Laboratory, USA) with a 512 multi-slice computed tomography (MSCT) scanner. The images were reconstructed with different strengths of the DLIR algorithm (i.e., low, medium, and high). SNR was measured by placing region of interest (ROI) in the center of CTP729 module (i.e. module for measuring CT number uniformity) of the Catphan phantom. CNR was measured by placing two ROIs in the center and a circular object with diameter of 15 mm on CTP730 module (i.e., module for measuring low-contrast detectability) of the Catphan phantom. Low contrast objects were also assessed visually on a 2D viewer software provided by Advantage Workstation Computer System (GE Healthcare, USA). The results show that the CT number for DLIR and FBP algorithms had no significant changes, but the noise levels on DLIR images were reduced compared to those from FBP. The noise levels were reduced to 2.9 HU (or 38.16%) for low-DLIR, 3.74 HU (or 49.21%) for medium-DLIR and 4.7 HU (or 61.84%) for high-DLIRh. Therefore, the SNR improved by 37.5, 49.35 and 61.94% for low-, medium-, and high-DLIR, respectively. The CNR were 0.68, 0.98, 1.2, 1.46 for FBP, low-DLIR, medium-DLIR and high-DLIR, respectively. Compared to CNR from FBP, the CNR improved by 30.61% for low-DLIR, 43.33% for medium-DLIR and 53.42% for high-DLIR. These results corresponded to our visual assessment that LCD is improved as the strength of the DLIR algorithm is increased. In conclusion, it is found that if the strength of the DLIR algorithm increases, then the noise level is reduced and the SNR and CNR are increased. Hence, the DLIR algorithm has a better conspicuity to detect low-contrast objects compared to those from FBP. [ABSTRACT FROM AUTHOR]

Early identification and diagnosis of COVID-19 are critical factors in relation to infection constraints. Computed tomography imaging is considered a golden modality for detecting and evaluating lung COVID-19 infection. In this paper, a currently proposed algorithm has been developed for the three-dimensional visualization of infection regions, where the segmentation algorithm combines a grayscale morphological reconstruction technique with a fast marching method, then reconstructing the stacked two-dimensional infection regions into a three-dimensional faction based on a strategy of marching cubes, the results showed the consistent determination of ground glass opacity infection region compared to the manual delineation of a senior physician, The suggested approach promotes robust medical image analysis based on sparse data and improves COVID-19 segmentation technology state of the art. [ABSTRACT FROM AUTHOR]

3D reconstruction is the inverse process of describing a 3D object or scene image and a computer technique that uses mathematical formulas to recover 3D information on the surface of an object from a 2D image. The internal cavity of the body observed by the surgeon through the endoscope during surgery is a flat figure, which cannot reflect the depth of the lesion, surrounding tissue, and other three-dimensional information. Endoscopic 3D reconstruction technology is clearer and more intuitive, helping doctors to operate on the lesion location more accurately, and is the technical basis for surgical navigation and virtual surgery. However, images taken by endoscopy generally have disadvantages such as imaging distortion and sparse organ tissue texture, increasing the difficulty of image feature extraction and matching. According to the number of cameras in monocular, binocular, and multi-vision, this paper investigates endoscopy-based 3D reconstruction techniques. The algorithms used in 3D reconstruction using binocular and multi-vision systems will take up more computing resources, making real-time 3D reconstruction more difficult. This paper analyzes these algorithms by comparing their algorithm principles, point cloud accuracy, sparsity, and response speed. And some possible solutions to the existing problems are proposed and future research directions are discussed. [ABSTRACT FROM AUTHOR]

Artificial Intelligence (AI) has been employed in medical imaging for many years to assist in various tasks, such as automated segmentation and classification. Advances in Artificial Neural Networks (ANNs), Machine Learning, the availability of larger data sets, and improved computing power are enabling the use of AI algorithms during image reconstruction in clinical CT scanners. In CT image reconstruction, these new algorithms can produce images with lower noise and improved image quality with the potential for further reduction in radiation dose and improved diagnostic confidence. This review summarizes CT image reconstruction from filtered backprojection (FBP) to recent AI approaches from three manufacturers. A variety of experiences gained during clinical implementation are shared, including phantom and observer studies. Recommendations include to optimize scan protocols first, then establish appropriate de-noising levels before reducing scan dose. Also important are awareness of a situation dubbed the "acutance trap" and the link between the performance of the neural networks and the data used to train them. [ABSTRACT FROM AUTHOR]

The normal Radon transform as well as the circular Radon transform are reviewed in this work. Traditionally, the first is linked to computerized tomography and the second is linked to synthetic aperture techniques that are well-established in both, radar, and ultrasound imaging systems. A new method that adapts the ordinary algebraic reconstruction methods to numerically reconstruct an image from its circular Radon transform is presented. To this end, a representation of an image using the unit box functions is employed, this representation facilitates the formation of an exact matrix of coefficients for a linear system to be solved. Regularized least square is used to solve the relatively small systems, and a machine learning approach is used for large systems. Several experiments are conducted for validation of the built system as well the reconstruction of images from their circular Radon transform. [ABSTRACT FROM AUTHOR]

The normal Radon transform as well as the circular Radon transform are reviewed in this work. Traditionally, the first is linked to computerized tomography and the second is linked to synthetic aperture techniques that are well-established in both, radar, and ultrasound imaging systems. A new method that adapts the ordinary algebraic reconstruction methods to numerically reconstruct an image from its circular Radon transform is presented. To this end, a representation of an image using the unit box functions is employed, this representation facilitates the formation of an exact matrix of coefficients for a linear system to be solved. Regularized least square is used to solve the relatively small systems, and a machine learning approach is used for large systems. Several experiments are conducted for validation of the built system as well the reconstruction of images from their circular Radon transform. [ABSTRACT FROM AUTHOR]

Computer vision is the core of recent research to extract useful information of the computer generated 2D images for solving controlling tasks in industrial applications. In vision research, representing an image in 2D will lose some depth information like geometrical relations between the objects. Disparity estimation has added the essence for the basic need of finding depth information. Initially, a survey was conducted on robotics based vision algorithms which revealed some issues such as correspondence, rectification and reconstruction that had not been properly analyzed. Subsequently, a few vision algorithms were developed to provide some partial solutions for the correspondence problem. These algorithms tried to minimize the computational complexity. Despite the attempts made by a few 3D stereo vision techniques for addressing these problems, there has been no achievement of satisfactory accuracy. The research work is to increase the accuracy of the 3D reconstruction with optimum level of mismatches. [ABSTRACT FROM AUTHOR]

A major field of research in low-dosage CT imaging is the development of image restoration techniques for computation tomography (CT), which considerably minimize patient radiation exposure while retaining high image quality. We suggest a new penalized low-square weighted (PWLS) application form, which will take advantage of regularisation based on the effective Transform Union. The square transformation union is pre-learned from a variety of picture patches derived from CT pictures and volumes. By switching between an Images acquired restructuring process and a small encoding and clustering point, the proposed PWLS-based cost function can be optimized. A relaxed explanation for these results augmented the Lagrangian process with ordered subsets, decreasing the number of forwarding and back projections used in the CT Image reconstruction phase. Simulation by 2D and 3D axial CT scans with the XCAT fantasy and 3D cylindrical chest, and abdominal scans demonstrate that the proposed technique greatly increases image quality with a non-adaptive edge-maintaining regularizer, both at standard and low dosage levels, relative to PWLS reconstruction. The regularisation of PWLS based on a community of learning transformations results in enhanced image restoration than using a standard taught rectangle. PWLS-ULTRA is also assisted by patches that increase image clarity and enhance the uniformity of image resolution. Compared with the over-complete dictionaries learned, but critically the method is much faster highly scalable. The methodology is equally or more effective. [ABSTRACT FROM AUTHOR]

In this work, we suggest and discuss several new algorithms for iterative image reconstruction in emission tomography. The new iterative formulas allow varying the rate of convergence and image corruption by adjusting the parameters of the algorithms. [ABSTRACT FROM AUTHOR]

In this work we developed and validated a Monte Carlo simulation framework of the imaging systems of the Varian TrueBeam STx Linac, and studied the scatter contribution in MV and kV cone-beam CT images. The simulations were validated by comparing measured PDD (Percentage Depth Dose) curves, off-axis profiles and transmission curves with those calculated with the DOSXYZnrc code and a dosimetric model of the detector response using the theoretical energy spectra. The transmission curves were obtained by imaging Al and Cu step-wedge phantoms with the MV system. A parallelized code was used to estimate cone-beam CT projections separating the scatter and primary contributions, optimizing the code using variance reduction techniques (VRTs). The models developed in this work show very good agreement with experimental measurements and with previous results reported in the literature. The use of VRTs allowed quick estimation of the scatter contributions with a speed up factor of 20. This MC framework was used to develop an iterative scatter correction method based on fast simulations of low-resolution voxel models and sparse angular sampling. The iterative scatter correction method showed very good results in both kV and MV cone-beam CT images, improving image quality by reducing the cupping artifact and improving the estimation of the CT numbers. [ABSTRACT FROM AUTHOR]

Maintain image quality is to conduct quality assurance routinely on CT scan. There are several important parameters in the assessment of image quality including CT number, uniformity and linearity, low contrast. The research method uses the variation of the parameters kVp and mA with restrictions on the use of parameters only in variations of 80, 100, 140 kVp with 192 mAs and 100, 200, 300 mAs with 120 kVp. previously considered the parameters of the Phillips Ingenuity CT scan application for the head examination routine and review as a reference. The results of the study for the CT number value between the recommended HU value and the measurement of the HU value performed have a value that is close to, so that for the CT number the CT scan is still in very good range, for the Uniformity HU value at the center of the phantom -0.854±0.402 and around the phantom (clockwise 12,3,6,9) -1,374±0,396, results are uniformity still within the tolerance range of±5 HU. Linearity HU value equation linear function with R2= 0.8692, indicating that a very strong correlation between the value of the attenuation coefficient HU. Quartile analysis of HU values for the kVp and mAs varies with the results of the second quartile value. Of the variations of mAs 100, 200 and 300 there were no significant differences in the values and tended to be the same, namely for 100 mAs Q2 values=92.04, 200 mAs Q2=93.62, 300 mAs Q2 values=92.05, while in the kVp variations there were significant differences i.e. 80 kVp Q2=99.12, 140 kVp Q2=89.59. So that the HU value is not affected by the mAs parameter, it can be said that there is no significant change in image quality from the contrast resolution; low contrast but it is influenced by the kVp parameter with a value of HU increase of about 36% when kV is increased by range of 60 kVp. [ABSTRACT FROM AUTHOR]

As medical image segmentation is the present day challenge and a demand for researchers, several attempts have been made in this field. Fuzzy c-means clustering (FCM) is an algorithm that is affected by noise, thus objective-function is generally initiated with local-spatial information to enhance the robustness. But it’s implementation often results in rise in time complexity, caused by a repetitive computation of the distance separating pixels of local-spatial neighbours and clustering-centres. In order to solve this problem, a Modified FCM (MFCM) method is proposed in this paper. Initially, Image enhancement is performed using Contrast Limited Adaptive Histogram Equalization (CLAHE) then, the local-spatial data of images is subsumed into the MFCM algorithm with introduction of morphological-reconstruction operation (morphological opening operation) to assure detail preservation and noise immunity of images. Thirdly, alteration of membership partition build on the distance between pixels enclosing local-spatial neighbours and centres of clusters is exchanged with local membership filtering. Algorithm is quite efficient as median-membership filtering is capable of improvising partition matrix s ystematically. Hence, we propose that the explained algorithm achieves better results in less time. [ABSTRACT FROM AUTHOR]

Magnetoacoustic Tomography (MAT) is one of tomographic methods that utilizes acoustic signals (ultrasonic) produced by the interaction of an external magnetic field with electrons in an object's material to produce an image that describes the distribution of electrical conductivity in the object's material. MAT combines good contrast from electrical impedance tomography techniques and good spatial resolution from sonography. In this research study, a MAT system has been designed, with a method of injecting electric current that varies with time, into the object material under the influence of an external static magnetic field, or often called Magnetoacoustic Tomography with Current Injection (MAT-CI). Because the electric current is injected from the outside of the sample to be reconstructed, the sample used as an object must be conductive. To investigate the ability of the system that has been built, this experiment used various types of material with various shapes as samples. Overall, the system created can acquire the ultrasonic signal generated by the sample and this system is able to reconstruct the test sample shape image with the help of k-Wave tomography reconstruction software, which is a Matlab-based magneto/photo acoustic tomography image reconstruction software. [ABSTRACT FROM AUTHOR]

The design of the CMS detector is optimized for muon measurements. The muon system consists of gas ionization detector technologies. Cathode Strip Chambers (CSC) with both tracking and triggering capabilities are installed in the forward region. The first stage of muon reconstruction uses information from individual muon chambers and is thus called local reconstruction, in contrast to a subsequent global reconstruction where the information from all detectors is combined. First, 2-dimensional spatial points (rechits) describing where a muon crosses the CSC layers are built from the electrical signals induced by the charged particle traversing the chamber. Next, from the reconstructed hits, straight-line track segments are built within each chamber. Local reconstruction becomes particularly challenging at high instantaneous luminosities, which are expected at the HL-LHC. The high rate of particles traversing the detectors leads to increased rate of spurious rechits and segments thus increasing the combinatorial backgrounds. In this respect, work on improving the current and developing new algorithms is essential and is in progress. This document presents the existing local reconstruction algorithms used in the CMS cathode strip chambers. Their performance as well as ongoing efforts towards HL-LHC improvements are discussed. [ABSTRACT FROM AUTHOR]

X-ray Computed Tomography (CT) is a powerful nondestructive evaluation (NDE) tool to characterize internal defects and flaws, regardless of surface conditions and sample materials. After data acquisition from a series of X-ray 2D projection imaging, reconstruction methods play a key role to convert raw data (2D radiography) to 3D models. For the past 50 years, standard reconstruction have been performed using analytical methods based on filtered back-projection (FBP) concepts. Numerous iterative methods that have been developed have shown some improvements on certain aspects of the reconstruction quality, but have not been widely adopted due to their high computational requirements. With modern high performance computing (HPC) and graphics processing unit (GPU) technologies, the computing power barrier for iterative methods have been reduced. Iterative methods have more potential to incorporate physical models and a priori knowledge to correct artifacts generated from analytical methods. In this work, we propose a generalized framework for iterative reconstruction with GPU acceleration, which can be adapted for different physical and statistical models in the inner iteration during reconstruction. The forward projection algorithm is an important part of the framework, and is analogous to the ray casting depth map algorithm that was implemented in an earlier work [I] and accelerated using the GPU. Within this framework, different sub-models could be developed in future to deal with different artifacts, such as beam hardening effect and limited angle data problem. [ABSTRACT FROM AUTHOR]

An inner tracking system (ITS) based on silicon pixel sensors is currently considered as one of the possible MPD upgrade steps. The main purpose of the new detector is to provide a better precision of the primary and secondary vertex reconstruction and improve track reconstruction in MPD in the region close to the interaction point. To study the ITS performance a new track finding algorithm was developed, which better takes into account the new system's advantages. In this paper the new algorithm is described and first results obtained on simulated data are presented. [ABSTRACT FROM AUTHOR]

A comparison between processing times of a standard serial algorithm and a parallel algorithm using Graphics Processing Unit (GPU) is presented here. Analysis was performed on an image of the Shepp–Logan phantom, obtained in a computed tomography (CT) study. Time, number of iterations and the structural similarity index (SSIM) metrics were use to quantify the quality of the reconstructed image by both methods. Parallel algorithms developed in CUDA can reconstruct images in times 2 orders of magnitude less than traditional methods, while keeping the same image quality. In conclusion, this system presents a good cost/effective setup for CT image reconstruction, viable for institutions with limited resources and countries of the developing world. [ABSTRACT FROM AUTHOR]

All commercial nuclear power plants (NPPs) in the United States contain concrete structures. These structures provide important foundation, support, shielding, and containment functions. Identification and management of aging and the degradation of concrete structures is fundamental to the proposed long-term operation of NPPs. Concrete structures in NPPs are often inaccessible and contain large volumes of massively thick concrete. While acoustic imaging using the synthetic aperture focusing technique (SAFT) works adequately well for thin specimens of concrete such as concrete transportation structures, enhancements are needed for heavily reinforced, thick concrete. We argue that image reconstruction quality for acoustic imaging in thick concrete could be improved with Model-Based Iterative Reconstruction (MBIR) techniques. MBIR works by designing a probabilistic model for the measurements (forward model) and a probabilistic model for the object (prior model). Both models are used to formulate an objective function (cost function). The final step in MBIR is to optimize the cost function. Previously, we have demonstrated a first implementation of MBIR for an ultrasonic transducer array system. The original forward model has been upgraded to account for direct arrival signal. Updates to the forward model will be documented and the new algorithm will be assessed with synthetic and empirical samples. [ABSTRACT FROM AUTHOR]

This paper proposes a computerized technique aimed at detecting heart beat positions in electrocardiograms (ECGs). The algorithm was implemented by adapting one of the most popular methods (Pan-Tompkins) developed to detect beat positions in ECG records, to be used in different leads of multi-channel ECGs. First, the Pan-Tompkins algorithm was programmed, and next, a lead-dependent version was implemented, adapting thesholds for each of the ECG channels. The original algorithm was tested with the MITBIH Arrhythmia Database, while the performance of the improved technique was tested on the different channels of records from the St. Petersburg Database. For the original Pan-Tompkins algorithm, a sensitivity of Se = 99.81% was obtained, with a positive predictivity (P+) = 99.85%. The F-Score was 0.9587, and the RMS RR Interval Error (RMSRRIE) resulted to be 4,480.46 ms. For the modified algorithm, results yielded an average value for Se = 99.92%, P+ = 99.98%, F-Score = 0.9718, and a mean value of 111.05 ms. for the RMSRRIE. Results indicate that our improved method provides similar better sensitivity and positive predictivity, increased F-Score, and it significantly reduces the temporal error when estimating the positions of QRS complexes (combination of Q, R and S waves representing ventricular depolarization in an ECG signal). [ABSTRACT FROM AUTHOR]

recent advances in microcopy and improvements in image processing algorithms have allowed the development of computer-assisted analytical approaches in cell identification. Several applications could be mentioned in this field: cellular phenotype identification, disease detection and treatment, identifying virus entry in cells and virus classification; these applications could help to complement the opinion of medical experts. Although many surveys have been presented in medical image analysis, they focus mainly in tissues and organs and none of the surveys about image cells consider an analysis following the stages in the typical image processing: segmentation, feature extraction and classification. The goal of this study is to provide comprehensive and critical analyses about the trends in each stage of cell image processing. In this paper we present a literature survey about cell identification using different image processing techniques. [ABSTRACT FROM AUTHOR]

Flare is a practical system for streaming 360-degree videos on smartphones. It takes a viewport-adaptive approach, which fetches only portions of a panoramic scene that cover what a viewer is about to perceive. Flare consists of a novel framework for the end-to-end streaming pipeline, introduces innovative streaming algorithms, and brings numerous system-level optimizations. In our demo, we will show that Flare substantially outperforms traditional viewport-agnostic streaming algorithms in terms of the video quality. We will also invite the audience to use Flare to watch attractive 360-degree videos. [ABSTRACT FROM AUTHOR]

The goal of this work was to design, develop and test practical solutions that could become the core of computer-aided design (CAD) software, simulating to meet the challenges of preoperative preparation in the shortest time. Its basis was adapted to clinical and technical requirements of the diagnostic image processing algorithm in cancer patients with the following construction of the model and manufacture of the replacement implant on the basis of its 3Dprototype. The results are useful for preoperative planning and production of anatomically accurate individual implants and endoprosthesis. [ABSTRACT FROM AUTHOR]

The large majority of industrial CT systems reconstruct the 3D volume by using an acquisitionon a circular trajec-tory. However, when inspecting long objects which are highly anisotropic, this scanning geometry creates severe artifacts in the reconstruction. For this reason, the use of an advanced CT scanning method like helical data acquisition is an efficient way to address this aspect known as the long-object problem. Recently, several analytically exact and quasiexact inversion formulas for helical cone-beam reconstruction have been proposed. Among them, we identified two algorithms of interest for our case. These algorithms are exact and of filtered back-projection structure. In this work we implemented the filtered-backprojection (FBP) and backprojection-filtration (BPF) algorithms of Zou and Pan (2004). For performance evaluation, we present a numerical compari-son of the two selected algorithms with the helical FDK algorithm using both complete (noiseless and noisy) and truncated data generated by CIVA (the simulation platform for non-destructive testing techniques developed at CEA). [ABSTRACT FROM AUTHOR]

Anew image compression algorithm is proposed and demonstrated in the context of structured light 3D reconstruction. Structured light images contain patterns of light, which are captured by the sensor at very high resolution. The algorithm steps involvea two level Discrete Wavelet Transformation (DWT) followed by a Discrete Cosine Transformation (DCT) to generate a DC-Column andan MA-Matrix (Multi-Array Matrix). The MA-Matrix is then partitioned into blocks and a minimization algorithm codes each block followed by arithmetic coding. At decompression stage anewproposedalgorithm, Sequential-Search Algorithm (SS-Algorithm) is used to estimate the MA-Matrix. Thereafter, all decompressed DC-Columns arecombined withthe MA-Matrix followed by inverse DCT and inverse DWT. The effectiveness of the algorithm is demonstrated within a 3D reconstruction scenario from structured light images. [ABSTRACT FROM AUTHOR]

A novel structured light method for color 3D surface profilometry is proposed. The proposed method may be used for reconstruction of both static and moving objects and does not require accurate color calibration of a camera-projector pair. The method is based on a structured light pattern that is a combination of a De Bruijn color sequence and of a sinusoidal fringe pattern. For static scenes phase-shifting and the De Bruijn window property are combined to obtain a high accuracy reconstruction. For dynamic scenes the Hessian ridge detector is applied to find stripes whose colors are decoded using a Gaussian mixture model. The De Bruijn window property and dynamic programming then uniquely identify projected stripes yielding a spatially sparser reconstruction. We have tested the proposed method on multiple objects with challenging surfaces and different albedos that clearly demonstrate both usability and robustness. [ABSTRACT FROM AUTHOR]

Block Compressed Sensing (BCS) is a new image sampling/compressing method with compressed sensing (CS). To solve the performance degradation of BCS-SPL (BCS with Smoothed Projected Landweber algorithm) at low sampling-ratio, we propose a novel algorithm called Total Variation based Adaptive-Sampling BCS with OMP (TVAS-BCS-OMP). TVAS-BCS-OMP blocks the whole image in an overlapping way to eliminate blocking effect. It assigns sampling-ratio depending on each block' texture complexity, which is measured by the block's Total Variation (TV) so that the blocks with big TV can attain higher sampling-ratio. Then only limited nonzero coefficients in each block are retained according to the adaptively assigned sampling-ratio. At last, we sample the blocks and conducts OMP reconstruction respectively. The experimental results show that under the condition of low initial sampling-ratio (lower than 0.2), TVAS-BCS-OMP achieves better reconstruction precision than BCS-SPL, especially in the blocks with complex texture. In addition, the new algorithm costs shorter reconstruction time than BCS-SPL algorithm. [ABSTRACT FROM AUTHOR]

The problem of automatically recognizing human faces from the facial images under varying expression and illumination is still attractive. The sparse signal representation is a new topic for object classification and face recognition. Implementing the sparse representation needs using a proper reconstruction algorithm to recover the sparse signal as well. In this paper, the performance of three greedy algorithms named matching pursuit (MP), orthogonal matching pursuit (OMP), compressive sampling matching pursuit (CoSaMP) and two l1 -minimization algorithms named primal-dual interior-point (PDIP) and Homotopy are compared according to achieved accuracy, consuming time, and the number of iterations. For this purpose, three well known datasets named, AR, Extended Yale B and Essex University are used. [ABSTRACT FROM AUTHOR]

Procedural modeling techniques can be used to encode a geometric shape on a high and abstract level: each class of objects and shapes is represented by one algorithm; and each artifact is one set of high-level parameters. In this paper, we use a generative object description and register it to real-world data e.g. laser scans. Afterwards, we can use the fitted procedural model to modify existing 3D shapes. The high-level description can be used to resemble realworld objects or create new ones. In this way, we can design shapes using both low-level details and high-level shape parameters at the same time. [ABSTRACT FROM AUTHOR]

In this work, we investigate the applicability of the constrained total-variation (TV)-minimization reconstruction method to industrial CT system. In general, industrial CT systems have the same principles of imaging process with clinical CT systems, but different imaging objectives and evaluation metrics. Optimization-based image reconstruction methods have been actively developed to meet practical challenges and extensively tested for clinical CT systems. However, the utility of optimization-based reconstruction methods is task-specific and not necessarily transferrable among different tasks. In this work, we adopt constrained TV-minimization programs together with adaptive-steepest-descent-projection-onto-convex-sets (ASD-POCS) algorithm for reconstructing images from data of a concrete sample collected using a laboratory industrial CT system developed for non-destructive evaluation. Our results, compared to those reconstructed from FBP-based algorithm, suggest that the constrained TV-minimization program combined with ASD-POCS algorithm can yield images with comparable or improved visual quality and achieve equivalent or better imaging objectives over [ABSTRACT FROM AUTHOR]

This paper presents results of a TFM implementation for Full Matrix Capture acquisitions in CIVA, proposed as a post-processing tool for accurate analysis. This implementation has been made on GPU architecture with OpenCL to minimize the processing time and offer computational device flexibility (GPU/CPU). Examples on immersion configurations on isotropic 2D CAD specimen with planar extrusion are proposed to illustrate the performances. Reconstructions on 2D or 3D areas of direct echoes with mode conversion are allowed. Probe scanning can also be taken into account. Reconstruction results and a benchmark explaining the speedup are presented. Further improvements are also reviewed. [ABSTRACT FROM AUTHOR]

In the nondestructive evaluation (NDE) of steel plates, besides the detection of cracks, what is also important is their characterization. Characterization is necessary for determining whether any discovered crack demands withdrawal of the part from service. In eddy current crack identification the response of a part to a coil is compared to the response without a crack. When different, the presence of the crack is flagged. But to characterize it, the computed response from eddy current analysis with a crack described by parameters is optimized to match measurements. This is heavy computation. Recently, Graphical Processing Unit (GPU) computing has had great success in many very large numerical computations. In this work we discuss the often undiscussed GPU memory limitation in Finite Element Optimization. In GPU computing the memory of the NVIDIA GPU is limited (4 GB on a PC today). This paper assesses the memory limits in terms of matrix size in light of the various ways to store a large matrix in order to overcome these limits. We revive old element-by-element finite element solvers from the early 1980s of working on a highly memory limited PC 282 to launch a GA kernel on thousands of CUDA threads exploiting the NVIDIA GPU architecture. [ABSTRACT FROM AUTHOR]

This paper presents a robust frequency diversity based algorithm for clutter reduction in ultrasonic A-scan waveforms. The performance of conventional spectral-temporal techniques like Split Spectrum Processing (SSP) is highly dependent on the parameter selection, especially when the signal to noise ratio (SNR) is low. Although spatial beamforming offers noise reduction with less sensitivity to parameter variation, phased array techniques are not always available. The proposed algorithm first selects an ascending series of frequency bands. A signal is reconstructed for each selected band in which a defect is present when all frequency components are in uniform sign. Combining all reconstructed signals through averaging gives a probability profile of potential defect position. To facilitate data collection and validate the proposed algorithm, Full Matrix Capture is applied on the austenitic steel and high nickel alloy (HNA) samples with 5MHz transducer arrays. When processing A-scan signals with unrefined parameters, the proposed algorithm enhances SNR by 20dB for both samples and consequently, defects are more visible in B-scan images created from the large amount of A-scan traces. Importantly, the proposed algorithm is considered robust, while SSP is shown to fail on the austenitic steel data and achieves less SNR enhancement on the HNA data. [ABSTRACT FROM AUTHOR]

A novel, autonomous reconfigurable robotic inspection system for quantitative NDE mapping is presented. The system consists of a fleet of wireless (802.11g) miniature robotic vehicles, each approximately 175 x 125 x 85 mm with magnetic wheels that enable them to inspect industrial structures such as storage tanks, chimneys and large diameter pipe work. The robots carry one of a number of payloads including a two channel MFL sensor, a 5 MHz dry coupled UT thickness wheel probe and a machine vision camera that images the surface. The system creates an NDE map of the structure overlaying results onto a 3D model in real time. The authors provide an overview of the robot design, data fusion algorithms (positioning and NDE) and visualization software. [ABSTRACT FROM AUTHOR]

In this paper we present an efficient implementation of an algorithm for reconstructing a 3D volume from limited angle projection data, based on statistical inversion theory. We demonstrate the strength of the method for detecting structural defects in large composite aerospace components, whose dimensions prevent acquiring measurements over the full circle. In comparison with a number of other tomographic reconstruction methods that can be applied to the limited angle case, such as tomosynthesis or simultaneous algebraic reconstruction technique (SART), we achieve superior depth resolution with reduced noise and artifacts. [ABSTRACT FROM AUTHOR]

A well-known damage detection paradigm is the use of ultrasonic guided waves that are generated and recorded by a spatially-distributed array of piezoelectric transducers. This type of configuration is capable of interrogating a defect from a variety of angles and over a large region of interest by analyzing all pairwise transducer signals. By subtracting prerecorded baseline signals, differential signals are obtained that can be analyzed for the purpose of detecting, locating, and characterizing newly-introduced scatterers. Typical analysis techniques such as delay-and-sum imaging have the ability to detect flaws, but their performance is limited, especially when the potential scatterers may have high directionality or introduce phase shifts. Signal envelopes are frequently used to avoid the problem of unknown phase shifts, which further reduces performance. The sparsity-based technique described here uses a different approach, where each potential damage location has its own generic linear scattering model that allows for unknown variations in amplitude and phase between each transducer pair. The differential signals are then assumed to be a linear combination of a small number of these models, and an image is generated using a block-sparse reconstruction algorithm that splits the signals into location-based components. Results are presented for experimental data. The images exhibit smaller spot size and fewer artifacts than those obtained via delay-and-sum imaging, provided the model is reasonably well-matched to the data. [ABSTRACT FROM AUTHOR]

A three-dimensional MPA probe has been incorporated with a high speed phased array electronic board to visualize nugget images of resistance spot welds. The primary application area of this battery operated portable MPA ultrasonic imaging system is in the automotive industry which a conventional destructive testing process is commonly adopted to check the quality of resistance spot welds in auto bodies. Considering an average of five-thousand spot welds in a medium size passenger vehicle, the amount of time and effort given to popping the welds and measuring nugget size are immeasurable in addition to the millions of dollars' worth of scrap metals recycled per plant per year. This wasteful labor intensive destructive testing process has become less reliable as auto body sheet metal has transitioned from thick and heavy mild steels to thin and light high strength steels. Consequently, the necessity of developing a non-destructive inspection methodology has become inevitable. In this paper, the fundamental aspects of the current 3-D probe design, data acquisition algorithms, and weld nugget imaging process are discussed. [ABSTRACT FROM AUTHOR]

In this paper the performance of total focusing method is compared with the widely used time-reversal MUSIC super resolution technique. The algorithms are tested with simulated and experimental ultrasonic array data, each containing different noise levels. The simulated time domain signals allow the effects of array geometry, frequency, scatterer location, scatterer size, scatterer separation and random noise to be carefully controlled. The performance of the imaging algorithms is evaluated in terms of resolution and sensitivity to random noise. It is shown that for the low noise situation, time-reversal MUSIC provides enhanced lateral resolution when compared to the total focusing method. However, for higher noise levels, the total focusing method shows robustness, whilst the performance of time-reversal MUSIC is significantly degraded. [ABSTRACT FROM AUTHOR]

Improving the ultrasound inspection capability for coarse grain metals remains of longstanding interest to industry and the NDE research community and is expected to become increasingly important for next generation power plants. A test sample of coarse grained Inconel 625 which is representative of future power plant components has been manufactured to test the detectability of different inspection techniques. Conventional ultrasonic A, B, and C-scans showed the sample to be extraordinarily difficult to inspect due to its scattering behaviour. However, in recent years, array probes and Full Matrix Capture (FMC) imaging algorithms, which extract the maximum amount of information possible, have unlocked exciting possibilities for improvements. This article proposes a robust methodology to evaluate the detection performance of imaging algorithms, applying this to three FMC imaging algorithms; Total Focusing Method (TFM), Phase Coherent Imaging (PCI), and Decomposition of the Time Reversal Operator with Multiple Scattering (DORT MSF). The methodology considers the statistics of detection, presenting the detection performance as Probability of Detection (POD) and probability of False Alarm (PFA). The data is captured in pulse-echo mode using 64 element array probes at centre frequencies of 1MHz and 5MHz. All three algorithms are shown to perform very similarly when comparing their flaw detection capabilities on this particular case. [ABSTRACT FROM AUTHOR]

In recent years, great advance has been made on electrostatic sensing technique for gas-solid flow measurement. Electrostatic tomography(EST) has been used in experiment researches as a novel non-intrusive measurement technique. Electrostatic sensor array is one of the key parts of electrostatic tomography system. Based on the image reconstruction algorithm, the charge on the particles can be obtained from the electrostatic measurement signals. However, reports on the relationship between the electrostatic signal acquired by the electrostatic sensor array and flow characteristics of the particles were very few. In this paper the mathematical model of the electrostatic sensor array was adopted, and its spatial sensitivity field was investigated. In the experiment, the electrostatic signals of the quarter flow and full pipe flow were acquired by the electrostatic sensor array. Based on the EST experiment and Power Spectrum Analysis of Signal, the energy distributions of those two flow patterns were compared. Results show that the sensitivity of the electrostatic sensor array is inhomogeneous in three-dimensional space. For this reason, the energy distributions of those two flow patterns are distinguished. [ABSTRACT FROM AUTHOR]

A Model Based Iterative Image Reconstruction (MoBIIR) algorithm is developed for impedance image reconstruction of Electrical Impedance Tomography (EIT). Simulated boundary data are generated for circular phantom configuration and the conductivity images are reconstructed with different regularization techniques and regularization parameters. Results show that the algorithm successfully reconstructs the conductivity distribution of the domain under test with its proper inhomogeneity and background conductivity. CNR and PCR of the reconstructed images are found high in both STR and LMR methods. Results show that the normalized projection errors in STR and LMR are same with higher value of λ where as the LMR method has low normalized projection errors compared to STR method with lower value of λ. [ABSTRACT FROM AUTHOR]