Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “algorithm selection”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 811 records · Page 45

Deep Reinforcement Learning based Routing in an Air-to-Air Ad-hoc Network

This paper studies the Multiple Sources and Multiple Destinations (MSMD) routing problem in a dynamic Air-to-Air Ad-hoc Network (AAAN). We consider a spectrum limited scenario where multiple links have to share the same frequency channel so that co-channel interference becomes inevitable. As a result, routing decisions and spectrum access are coupled and must be jointly considered. This paper proposes a deep Q-learning based algorithm to find an optimal routing and channel selection strategy that minimizes the end-to-end communication delay. Specifically, under the assumption that only local information is available to every node, the Deep Q-Network (DQN) is trained offline to learn the optimal routing and channel selection strategy. After the trained DQN is implemented in every node, multiple relay nodes can simultaneously determine their next-hop relay and channel selections in real-time. Simulation results demonstrate the efficacy of our proposed algorithm.

AAAN↗

Deep Reinforcement Learning based Routing in an Air-to-Air Ad-hoc Network

This paper studies the Multiple Sources and Multiple Destinations (MSMD) routing problem in a dynamic Air-to-Air Ad-hoc Network (AAAN). We consider a spectrum limited scenario where multiple links have to share the same frequency channel so that co-channel interference becomes inevitable. As a result, routing decisions and spectrum access are coupled and must be jointly considered. This paper proposes a deep Q-learning based algorithm to find an optimal routing and channel selection strategy that minimizes the end-to-end communication delay. Specifically, under the assumption that only local information is available to every node, the Deep Q-Network (DQN) is trained offline to learn the optimal routing and channel selection strategy. After the trained DQN is implemented in every node, multiple relay nodes can simultaneously determine their next-hop relay and channel selections in real-time. Simulation results demonstrate the efficacy of our proposed algorithm.

AAAN↗

Image Navigation and Registration Performance Assessment Tool Set for the GOES-R Advanced Baseline Imager and Geostationary Lightning Mapper

The GOES-R Flight Project has developed an Image Navigation and Registration (INR) Performance Assessment Tool Set (IPATS) for measuring Advanced Baseline Imager (ABI) and Geostationary Lightning Mapper (GLM) INR performance metrics in the post-launch period for performance evaluation and long term monitoring. For ABI, these metrics are the 3-sigma errors in navigation (NAV), channel-to-channel registration (CCR), frame-to-frame registration (FFR), swath-to-swath registration (SSR), and within frame registration (WIFR) for the Level 1B image products. For GLM, the single metric of interest is the 3-sigma error in the navigation of background images (GLM NAV) used by the system to navigate lightning strikes. 3-sigma errors are estimates of the 99.73rd percentile of the errors accumulated over a 24-hour data collection period. IPATS utilizes a modular algorithmic design to allow user selection of data processing sequences optimized for generation of each INR metric. This novel modular approach minimizes duplication of common processing elements, thereby maximizing code efficiency and speed. Fast processing is essential given the large number of sub-image registrations required to generate INR metrics for the many images produced over a 24-hour evaluation period. Another aspect of the IPATS design that vastly reduces execution time is the off-line propagation of Landsat based truth images to the fixed grid coordinates system for each of the three GOES-R satellite locations, operational East and West and initial checkout locations. This paper describes the algorithmic design and implementation of IPATS and provides preliminary test results.

Image Navigation↗

Image Navigation and Registration (INR) Performance Assessment Tool Set (IPATS) for the GOES-R Advanced Baseline Imager and Geostationary Lightning Mapper

The GOES-R Flight Project has developed an Image Navigation and Registration (INR) Performance Assessment Tool Set (IPATS) for measuring Advanced Baseline Imager (ABI) and Geostationary Lightning Mapper (GLM) INR performance metrics in the post-launch period for performance evaluation and long term monitoring. For ABI, these metrics are the 3-sigma errors in navigation (NAV), channel-to-channel registration (CCR), frame-to-frame registration (FFR), swath-to-swath registration (SSR), and within frame registration (WIFR) for the Level 1B image products. For GLM, the single metric of interest is the 3-sigma error in the navigation of background images (GLM NAV) used by the system to navigate lightning strikes. 3-sigma errors are estimates of the 99.73rd percentile of the errors accumulated over a 24 hour data collection period. IPATS utilizes a modular algorithmic design to allow user selection of data processing sequences optimized for generation of each INR metric. This novel modular approach minimizes duplication of common processing elements, thereby maximizing code efficiency and speed. Fast processing is essential given the large number of sub-image registrations required to generate INR metrics for the many images produced over a 24 hour evaluation period. Another aspect of the IPATS design that vastly reduces execution time is the off-line propagation of Landsat based truth images to the fixed grid coordinates system for each of the three GOES-R satellite locations, operational East and West and initial checkout locations. This paper describes the algorithmic design and implementation of IPATS and provides preliminary test results.

Image Registration↗

The STRong lensing Insights into the Dark Energy Survey (STRIDES) 2016 follow-up campaign – I. Overview and classification of candidates selected by two techniques

Here, the primary goals of the STRong lensing Insights into the Dark Energy Survey (STRIDES) collaboration are to measure the dark energy equation of state parameter and the free streaming length of dark matter. To this aim, STRIDES is discovering strongly lensed quasars in the imaging data of the Dark Energy Survey and following them up to measure time delays, high resolution imaging, and spectroscopy sufficient to construct accurate lens models. In this paper, we first present forecasts for STRIDES. Then, we describe the STRIDES classification scheme, and give an overview of the Fall 2016 follow-up campaign. We continue by detailing the results of two selection methods, the Outlier Selection Technique and a morphological algorithm, and presenting lens models of a system, which could possibly be a lensed quasar in an unusual configuration. We conclude with the summary statistics of the Fall 2016 campaign. Including searches presented in companion papers (Anguita et al.; Ostrovski et al.), STRIDES followed up 117 targets identifying 7 new strongly lensed systems, and 7 nearly identical quasars (NIQs), which could be confirmed as lenses by the detection of the lens galaxy. 76 candidates were rejected and 27 remain otherwise inconclusive, for a success rate in the range 6-35\%. This rate is comparable to that of previous searches like SQLS even though the parent dataset of STRIDES is purely photometric and our selection of candidates cannot rely on spectroscopic information.

79 ASTRONOMY AND ASTROPHYSICS↗

Marine Boundary Layer Cloud Property Retrievals from High-Resolution ASTER Observations: Case Studies and Comparison with Terra MODIS

A research-level retrieval algorithm for cloud optical and microphysical properties is developed for the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) aboard the Terra satellite. It is based on the operational MODIS algorithm. This paper documents the technical details of this algorithm and evaluates the retrievals for selected marine boundary layer cloud scenes through comparisons with the operational MODIS Data Collection 6 (C6) cloud product. The newly developed, ASTERspecific cloud masking algorithm is evaluated through comparison with an independent algorithm reported in Zhao and Di Girolamo (2006). To validate and evaluate the cloud optical thickness (tau) and cloud effective radius (r(sub eff)) from ASTER, the high-spatial-resolution ASTER observations are first aggregated to the same 1000m resolution as MODIS. Subsequently, tau(sub aA) and r(sub eff, aA) retrieved from the aggregated ASTER radiances are compared with the collocated MODIS retrievals. For overcast pixels, the two data sets agree very well with Pearson's product-moment correlation coefficients of R greater than 0.970. However, for partially cloudy pixels there are significant differences between r(sub eff, aA) and the MODIS results which can exceed 10 micrometers. Moreover, it is shown that the numerous delicate cloud structures in the example marine boundary layer scenes, resolved by the high-resolution ASTER retrievals, are smoothed by the MODIS observations. The overall good agreement between the research-level ASTER results and the operational MODIS C6 products proves the feasibility of MODIS-like retrievals from ASTER reflectance measurements and provides the basis for future studies concerning the scale dependency of satellite observations and three-dimensional radiative effects.

marine boundary layer↗

CDR3α drives selection of the immunodominant Epstein Barr virus (EBV) BRLF1-specific CD8 T cell receptor repertoire in primary infection

The T cell receptor (TCR) repertoire is an essential component of the CD8 T-cell immune response. Here, we seek to investigate factors that drive selection of TCR repertoires specific to the HLA-A2-restricted immunodominant epitope BRLF1 109-117 (YVLDHLIVV) over the course of primary Epstein Barr virus (EBV) infection. Using single-cell paired TCRαβ sequencing of tetramer sorted CD8 T cells ex vivo , we show at the clonal level that recognition of the HLA-A2-restricted BRLF1 (YVL-BR, BRLF-1109) epitope is mainly driven by the TCRα chain. For the first time, we identify a CDR3α (complementarity determining region 3 α) motif, KDTDKL, resulting from an obligate AV8.1-AJ34 pairing that was shared by all four individuals studied. This observation coupled with the fact that this public AV8.1-KDTDKL-AJ34 TCR pairs with multiple different TCRβ chains within the same donor (median 4; range: 1–9), suggests that there are some unique structural features of the interaction between the YVL-BR/MHC and the AV8.1-KDTDKL-AJ34 TCR that leads to this high level of selection. Newly developed TCR motif algorithms identified a lysine at position 1 of the CDR3α motif that is highly conserved and likely important for antigen recognition. Crystal structure analysis of the YVL-BR/HLA-A2 complex revealed that the MHC-bound peptide bulges at position 4, exposing a negatively charged aspartic acid that may interact with the positively charged lysine of CDR3α. TCR cloning and site-directed mutagenesis of the CDR3α lysine ablated YVL-BR-tetramer staining and substantially reduced CD69 upregulation on TCR mutant-transduced cells following antigen-specific stimulation. Reduced activation of T cells expressing this CDR3 motif was also observed following exposure to mutated (D4A) peptide. In summary, we show that a highly public TCR repertoire to an immunodominant epitope of a common human virus is almost completely selected on the basis of CDR3α and provide a likely structural basis for the selection. These studies emphasize the importance of examining TCRα, as well as TCRβ, in understanding the CD8 T cell receptor repertoire.

Microbiology↗

A balanced submatrix merging algorithm for multiprocessor architectures

In this article, a parallel algorithm which applies Givens rotations to selectively annihilate k(k + 1)/2 nonzero elements from two k x n(k not more than n) upper trapezoidal submatrices is described. The new algorithm is suitable for implementation on either a pair of directly connected local-memory processors or two clusters of multiple tightly-coupled processors. Analyses show that in both cases the proposed algorithms achieve optimal speed-up by balancing the work load distribution and masking interprocessor or intercluster communication by computation if k is much small than n. In the context of solving large scale least squares problems, this submatrix merging step is repetitively needed during the entire computation and, furthermore, there are usually many pairs of such submatrices to be merged with each submatrix stored in the memory of a processor or a cluster of processors. The proposed algorithm can be applied to each pair of submatrices concurrently, and thus parallelizes an important step in solving the least squares problems.

Chu, Eleanor↗

Reduce-Order Modeling of Multigroup Neutron Cross Sections for High-Temperature Gas-cooled Reactors

Deterministic neutronics calculations rely on multigroup neutron cross section libraries, which usually consists of a database of tabulated values, used to calculate the cross sections through multivariate linear interpolation. However, interpolation of the multidimensional cross section data becomes memory inefficient and time consuming as the number of tabulations increases, significantly slowing down the neutronics calculation, especially in the case of micro cross section libraries where every isotope (on the order of hundreds) has its own set of specific reactions and cross sections. To address this challenge, this work constructs efficient and robust reduced-order models (ROMs) of the multi-group cross sections to support the Griffin simulation of high-temperature gas-cooled reactors (HTGRs). The first part of the study investigates the linearity of the multi-group cross section data across isotopes, reaction types and energy groups on pre-generated datasets for the purpose of dimensionality reduction. Secondly, a down-selection of ROM techniques is presented on representative classical machine learning (ML) techniques, including variants of linear regression, kernel-based methods, tree-based algorithms, and artificial neural networks. The selection criteria jointly consider the memory efficiency, predictive accuracy, prediction speed, and scalability in comparison to the multidimensional interpolation. Among all the ML techniques, deep neural networks (DNNs) have proven to be the best selection with sufficient accuracy, high robustness, good memory efficiency, great scalability, and superior flexibility. DNNs for have been trained for all isotopes in this work and systematic Griffin testing is ongoing at this moment to ensure the feasibility of this ROM technique for cross section predictions.

42 - ENGINEERING↗

Reduced-Order Modeling of Multigroup Neutron Cross Sections for High-Temperature Gas-cooled Reactors

Abstract – Deterministic neutronics calculations rely on multigroup neutron cross section libraries, which consist of databases of tabulated values, used to calculate the neutron cross sections through multivariate linear interpolation. However, interpolation of the multidimensional cross section data becomes memory inefficient and time consuming as the number of tabulations increases, significantly slowing down the neutronics calculation, especially in the case of microscopic cross section libraries where every isotope (on the order of hundreds) has its own set of specific reactions and cross sections. In order to address this challenge, this work constructs efficient and robust reduced-order models (ROMs) of the multi-group cross sections to support the Griffin simulation of high-temperature gas-cooled reactors (HTGRs). The first part of the study investigates the linearity of the multigroup cross section data across isotopes, reaction types, and energy groups on pre-generated datasets for the purpose of dimensionality reduction. Secondly, a down-selection of ROM techniques is presented on representative classical machine learning (ML) techniques, including variants of linear regression, kernel-based methods, tree-based algorithms, and artificial neural networks. The selection criteria jointly consider the memory efficiency, predictive accuracy, prediction speed, and scalability in comparison to the multidimensional interpolation. Among all the ML techniques, deep neural networks (DNNs) have proven to be the best selection with sufficient accuracy, high robustness, good memory efficiency, great scalability, and superior flexibility. DNNs have been trained for all isotopes in this work and systematic Griffin testing is ongoing to ensure the feasibility of this ROM technique for predicting cross section and reducing memory requirements without a significant sacrifice in computational performance.

42 - ENGINEERING↗

Advanced Cross Section Library Generation using Reduced Order Models

Deterministic neutronics calculations rely on multigroup neutron cross section libraries, which consist of databases of tabulated values, used to calculate the neutron cross sections through multivariate linear interpolation. However, interpolation of the multidimensional cross section data becomes memory inefficient and time consuming as the number of tabulations increases, significantly slowing down the neutronics calculation, especially in the case of microscopic cross section libraries where every isotope (on the order of hundreds) has its own set of specific reactions and cross sections. In order to address this challenge, this work constructs efficient and robust reduced-order models (ROMs) of the multi-group cross sections to support the Griffin simulation of high-temperature gas-cooled reactors (HTGRs). The first part of the study investigates the linearity of the multigroup cross section data across isotopes, reaction types, and energy groups on pre-generated datasets for the purpose of dimensionality reduction. Secondly, a down-selection of ROM techniques is presented on representative classical machine learning (ML) techniques, including variants of linear regression, kernel-based methods, tree-based algorithms, and artificial neural networks. The selection criteria jointly consider the memory efficiency, predictive accuracy, prediction speed, and scalability in comparison to the multidimensional interpolation. Among all the ML techniques, deep neural networks (DNNs) have proven to be the best selection with sufficient accuracy, high robustness, good memory efficiency, great scalability, and superior flexibility. DNNs have been trained for all isotopes in this work and systematic Griffin testing is ongoing to ensure the feasibility of this ROM technique for predicting cross section and reducing memory requirements without a significant sacrifice in computational performance.

42 - ENGINEERING↗

Estimation of LOCA break size using cascaded support vector regression - 334

A failure of emergency core cooling system (ECCS) during a certain small break loss of coolant accident (SBLOCA) can cause a severe core uncovery and fuel failure. The safety injection systems (SISs) might not function properly in case of a SBLOCA, due to the slight change of pressure in the pipes. Early trend of the major parameters has to be observed and the accurate information has to be given to the nuclear power plant (NPP) operators by precisely identifying the break size to effectively manage accidents. In this study, the objective is to provide to operators the information on the break size in a short time by considering the accident situations such as hot-leg LOCA, cold-leg LOCA, steam generator tube rupture (SGTR). A cascaded support vector regression (CSVR) is used in order to estimate the break size. The simulation data set was obtained from the optimized power plant reactor 1000 (OPR1000) using modular accident analysis program (MAAP) code. And genetic algorithm (GA) was used to select the input variables of the CSVR model and optimize its parameters. As a result of this study, the CSVR model estimate very well the break size of LOCAs. If the operators can predict the break size in the LOCA, they can response quickly and properly to LOCA circumstances to prevent the core uncovery and fuel failure. Also, it will be possible to more efficiently manage beyond design basis accidents. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Uncertainty Quantification in CO2 Trapping Mechanisms: A Case Study of PUNQ-S3 Reservoir Model Using Representative Geological Realizations and Unsupervised Machine Learning

Evaluating uncertainty in CO2 injection projections often requires numerous high-resolution geological realizations (GRs) which, although effective, are computationally demanding. This study proposes the use of representative geological realizations (RGRs) as an efficient approach to capture the uncertainty range of the full set while reducing computational costs. A predetermined number of RGRs is selected using an integrated unsupervised machine learning (UML) framework, which includes Euclidean distance measurement, multidimensional scaling (MDS), and a deterministic K-means (DK-means) clustering algorithm. In the context of the intricate 3D aquifer CO2 storage model, PUNQ-S3, these algorithms are utilized. The UML methodology selects five RGRs from a pool of 25 possibilities (20% of the total), taking into account the reservoir quality index (RQI) as a static parameter of the reservoir. To determine the credibility of these RGRs, their simulation results are scrutinized through the application of the Kolmogorov–Smirnov (KS) test, which analyzes the distribution of the output. In this assessment, 40 CO2 injection wells cover the entire reservoir alongside the full set. The end-point simulation results indicate that the CO2 structural, residual, and solubility trapping within the RGRs and full set follow the same distribution. Simulating five RGRs alongside the full set of 25 GRs over 200 years, involving 10 years of CO2 injection, reveals consistently similar trapping distribution patterns, with an average value of Dmax of 0.21 remaining lower than Dcritical (0.66). Using this methodology, computational expenses related to scenario testing and development planning for CO2 storage reservoirs in the presence of geological uncertainties can be substantially reduced.

Mahjour, Seyed Kourosh↗

Comparative analysis of 11 different radioisotopes for palliative treatment of bone metastases by computational methods

Purpose: Throughout the years, the palliative treatment of bone metastases using bone seeking radiotracers has been part of the therapeutic resources used in oncology, but the choice of which bone seeking agent to use is not consensual across sites and limited data are available comparing the characteristics of each radioisotope. Computational simulation is a simple and practical method to study and to compare a variety of radioisotopes for different medical applications, including the palliative treatment of bone metastases. This study aims to evaluate and compare 11 different radioisotopes currently in use or under research for the palliative treatment of bone metastases using computational methods. Methods: Computational models were used to estimate the percentage of deoxyribonucleic acid (DNA) damage (fast Monte Carlo damage algorithm), the probability of correct DNA repair (Monte Carlo excision repair algorithm), and the radiation-induced cellular effects (virtual cell radiobiology algorithm) post-irradiation with selected particles emitted by phosphorus-32 ({sup 32}P), strontium-89 ({sup 89}Sr), yttrium-90 ({sup 90}Y ), tin-117 ({sup 117m}Sn), samarium-153 ({sup 153}Sm), holmium-166 ({sup 166}Ho), thulium-170 ({sup 170}Tm), lutetium-177 ({sup 177}Lu), rhenium-186 ({sup 186}Re), rhenium-188 ({sup 188}Re), and radium-223 ({sup 223}Ra). Results: {sup 223}Ra alpha particles, {sup 177}Lu beta minus particles, and {sup 170}Tm beta minus particles induced the highest cell death of all investigated particles and radioisotopes. The cell survival fraction measured post-irradiation with beta minus particles emitted by {sup 89}Sr and {sup 153}Sm, two of the most frequently used radionuclides in the palliative treatment of bone metastases in clinical routine practice, was higher than {sup 177}Lu beta minus particles and {sup 223}Ra alpha particles. Conclusions: {sup 223}Ra and {sup 177}Lu hold the highest potential for palliative treatment of bone metastases of all radioisotopes compared in this study. Data reported here may prompt future in vitro and in vivo experiments comparing different radionuclides for palliative treatment of bone metastases, raise the need for the careful rethinking of the current widespread clinical use of {sup 89}Sr and {sup 153}Sm, and perhaps strengthen the use of {sup 223}Ra and {sup 177}Lu in the palliative treatment of bone metastases.

60 APPLIED LIFE SCIENCES↗

Selective robust optimization: A new intensity-modulated proton therapy optimization strategy

Purpose: To develop a new robust optimization strategy for intensity-modulated proton therapy as an important step in translating robust proton treatment planning from research to clinical applications. Methods: In selective robust optimization, a worst-case-based robust optimization algorithm is extended, and terms of the objective function are selectively computed from either the worst-case dose or the nominal dose. Two lung cancer cases and one head and neck cancer case were used to demonstrate the practical significance of the proposed robust planning strategy. The lung cancer cases had minimal tumor motion less than 5 mm, and, for the demonstration of the methodology, are assumed to be static. Results: Selective robust optimization achieved robust clinical target volume (CTV) coverage and at the same time increased nominal planning target volume coverage to 95.8%, compared to the 84.6% coverage achieved with CTV-based robust optimization in one of the lung cases. In the other lung case, the maximum dose in selective robust optimization was lowered from a dose of 131.3% in the CTV-based robust optimization to 113.6%. Selective robust optimization provided robust CTV coverage in the head and neck case, and at the same time improved controls over isodose distribution so that clinical requirements may be readily met. Conclusions: Selective robust optimization may provide the flexibility and capability necessary for meeting various clinical requirements in addition to achieving the required plan robustness in practical proton treatment planning settings.

60 APPLIED LIFE SCIENCES↗

Geoscience Laser Altimeter System: Characteristics and Performance of the Altimeter Receiver

The Geoscience Laser Altimeter System (GLAS) on board ICESat spacecraft measures the surface height (altimetry) via the time of flight of its 1064 nm laser pulse. The GLAS laser transmitter produces 6 ns wide pulses with 70 mJ energy at 1064 nm at a 40 Hz rate. The altimeter receiver consists of a telescope, aft optics, a silicon avalanche photodiode, and electronic amplifiers. The transmitted and echo pulse waveforms are digitized at 1 GHz rate. The laser pulse time of flight is determined on the ground from the two digitized pulse waveforms and their positions in the full waveform record (about 5.4 ms ong) by computing the pulse centroids or by curve fitting. The GLAS receiver algorithms in on board software selects the two waveform segments containing the transmitted and the echo pulses and sends them to ground. The probability of echo pulse detection and the accuracy of time of flight measurement depend on the received signal level, the background light within the receiver field of view, the inherent detector and amplifier noise, the quantization of the digitizer, and some times by cloud obscurations. A receiver model has been developed to calculate the probability of detection and accuracy of the altimeter measurements with these noise sources. From prelaunch testing, the minimum detectable echo pulse energy for 90% detection probability was about 0.1 fj/pulse onto the detector. Such a receiver sensitivity allows GLAS to measure the surface height through clouds with optical density less than 2. The echo pulse energy required to achieve 10 cm ranging accuracy was found to be about 3 times higher than the minimum detectable signal level. The smallest single shot range measurement error, which was determined by ranging to a fixed target with strong echo pulses and no background light, was 2 to 3cm. The maximum linear response echo pulse energy was 10 fJ/pulse for the strongest echo signals, assuming a Lambertian scattering snow surface, clear sky atmosphere transmission, and no pulse width spreading due to surface slopes. Results from the initial 38 days of GLAS operation on-orbit with Laser 1, showed the echo pulse energies varied, as expected, over a wide dynamic range due to the variability of atmosphere transmission and Earth surface characteristics. The receiver was able to reliably detect the ground surface echoes and measure time of flight under such dynamic conditions, except when there were dense clouds. When over Antarctica, the echo pulse energies were several times stronger than predicted. This suggests that the 1064 nm backscattering from the snow pack and ice-sheet &ce are not completely Lambertian but is somewhat peaked at opposit.ion. With Laser 1, the peak amplitude of the echo pulses from flat ice sheets under clear sky conditions exceeded the linear response range of the receiver, causing some pulse waveform distortion due to saturation. We have characterized the effects of receiver saturation on the time of flight, pulse width, and pulse energy measurements for flat surface by testing a night spare detector in the lab with simulated echo pulses. A data processing algorithm that minimizes the errors due to saturation for these measurements are described.

Sun, Xiao-Li↗

Feasibility of an image planning system for kilovoltage image-guided radiation therapy

Purpose: Image guidance has become a standard of care for many treatment scenarios in radiation therapy. This is most typically accomplished by use of kV x-ray devices mounted onto the linear accelerator (Linac) gantry that yield planar, fluoroscopic, and cone-beam computed tomography (CBCT) images. Image acquisition parameters are chosen via preset techniques that rely on broad categorizations in patient anatomy and imaging goal. However, the optimal imaging technique results in detectability of the features of interest while exposing the patient to minimum dose. Herein, the authors present an investigation into the feasibility of developing an image planning system (IPS) for radiotherapy.Methods: In this first phase, the authors focused on developing an algorithm to predict tissue contrast produced by a common radiotherapy planar imaging chain. Input parameters include a CT dataset and simulated planar imaging technique settings that include kV and mAs. Energy-specific attenuation through each voxel of the CT dataset was calculated in the algorithm to derive a net transmitted intensity. The response of the flat panel detector was integrated into the image simulation algorithm. Verification was conducted by comparing simulated and measured images using four phantoms. Comparisons were made in both high and low contrast settings, as well as changes in the geometric appearance due to image saturation. Results: The authors studied a lung nodule test object to assess the planning system's ability to predict object contrast and detectability. Verification demonstrated that the slope of the pixel intensities is similar, the presence of the nodule is evident, and image saturation at high mAs values is evident in both images. The appearance of the lung nodule is a function of the image detector saturation. The authors assessed the dimensions of the lung nodule in measured and simulated images. Good quantitative agreement affirmed the algorithm's predictive capabilities. The invariance of contrast with kVp and mAs prior to saturation was predicted, as well as the gradual loss of object detectability as saturation was approached. Small changes in soft tissue density were studied using a mammography step wedge phantom. Data were acquired at beam qualities of 80 and 120 kVp and over exposure values ranging from 0.04 to 500 mAs. The data showed good agreement in terms of the absolute value of pixel intensities predicted, as well as small variations across the step wedge pattern. The saturation pixel intensity was consistent between the two beam qualities studied. Boney tissue contrast was assessed using two abdominal phantoms. Measured and calculated values agree in terms of predicting the mAs value at which detector saturation, and subsequent loss of contrast occurs. The lack of variation in contrast over mAs values lower than 10 suggests that there is wide latitude for minimizing patient dose. Conclusions: The authors developed and tested an algorithm that can be used to assist in kV imaging technique selection during localization for radiotherapy. Phantom testing demonstrated the algorithm's predictive accuracy for both low and high contrast imaging scenarios. Detector saturation with subsequent loss of imaging detail, both in terms of object size and contrast were accurately predicted by the algorithm.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Efficient demultiplexing algorithm for noncontiguous carriers

A channel separation algorithm for the frequency division multiple access/time division multiplexing (FDMA/TDM) scheme is presented. It is shown that implementation using this algorithm can be more effective than the fast Fourier transform (FFT) algorithm when only a small number of carriers need to be selected from many, such as satellite Earth terminals. The algorithm is based on polyphase filtering followed by application of a generalized Walsh-Hadamard transform (GWHT). Comparison of the transform technique used in this algorithm with discrete Fourier transform (DFT) and FFT is given. Estimates of the computational rates and power requirements to implement this system are also given.

Thanawala, A. A.↗