Search NASA⌕ Search

SEARCH · Search NASA

Results for “Signal processing algorithms”

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 487 records · Page 27

A Computationally Efficient Algorithm for Computing Convex Hull Prices

Electricity markets worldwide allow participants to bid non-convex production offers. While non-convex offers can more accurately reflect a resource's capabilities, they create challenges for market clearing processes. For example, system operators may execute side payments when a participant’s cost is not covered through energy sale settlements from locational marginal pricing schemes, or when a participant incurs lost opportunity costs to follow the dispatch signal. Convex hull pricing minimizes these and other types of side payments while providing uniform (i.e., locationally and temporally consistent) prices. However, computing convex hull prices involves solving either a large-scale linear program - which in turn requires explicit descriptions of market participants’ convex hulls -or the Lagrangian dual of the corresponding non-convex scheduling problem. Here, we propose a computationally feasible and industrially scalable Benders decomposition approach to computing convex hull prices at least an order of magnitude faster than the current state-of-the-art while leveraging recent advances in convex hull formulations for thermal generating units.

61 RADIATION PROTECTION AND DOSIMETRY↗

Smart connected worker edge platform for smart manufacturing: Part 2—Implementation and on‐site deployment case study

Abstract In this paper, we describe specific deployments of the Smart Connected Worker (SCW) Edge Platform for Smart Manufacturing through implementation of four instructive real‐world use cases that illustrate the role of people in a Smart Manufacturing paradigm through which affordable, scalable, accessible, and portable (ASAP) information technology (IT) acquires and contextualizes data into information for transmission to operation technologies (OT). For case one, the platform captures the relationships between energy consumption and human workflows for improved energy productivity while workers interact with machines during semiconductor manufacturing. The platform utilizes human cognition to identify anomalous machine behavior for root cause analysis of system faults via neural network (NN) that recognize alarm postures of workers with cameras. For case two, a smart assembly line is demonstrated for state monitoring and fault detection. Machine learning (ML) models are used to recognize system states and identify fault scenarios with human intervention. For case three, the platform monitors human–machine interactions to classify manufacturing machine states for proper operations and energy productivity. Internal energy states of individual or collections of manufacturing equipment are determined via NN based algorithms that disaggregate signals associated with smart metering typically deployed at manufacturing facilities. These methods predict the real time energy profile of each machine from the total energy profile of a manufacturing site. For case four, a software defined sensor system built with scientific workflow engines is demonstrated for contextualizing data from laser surface refraction for characterization, and diagnostics in the processing of additively manufactured titanium alloy.

Donovan, Richard P.↗

Optimal analysis of azimuthal features in the CMB

We present algorithms for searching for azimuthally symmetric features in CMB data. Our algorithms are fully optimal for masked all-sky data with inhomogeneous noise, computationally fast, simple to implement, and make no approximations. We show how to implement the optimal analysis in both Bayesian and frequentist cases. In the Bayesian case, our algorithm for evaluating the posterior likelihood is so fast that we can do a brute-force search over parameter space, rather than using a Monte Carlo Markov chain. Our motivating example is searching for bubble collisions, a pre-inflationary signal which can be generated if multiple tunneling events occur in an eternally inflating spacetime, but our algorithms are general and should be useful in other contexts.

79 ASTRONOMY AND ASTROPHYSICS↗

RTN-117: Image Calibration and Instrument Signal Removal for the First Year of the LSST

The NSF-DOE Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST) requires calibration products that provide uniform, stable, and accurate photometric and astrometric performance across the 3.2 gigapixel focal plane and throughout the 10-year survey. This paper details the algorithms and workflows used to produce instrument calibrations and remove instrumental artifacts for Data Preview 2 (DP2)---the first end-to-end processing demonstration using on-sky data with the LSST Camera (LSSTCam). We describe the verification, acceptance, and certification framework used to assess calibration quality and quantify residual systematics. We show baseline metrics on calibrated science images to evaluate the robustness of the calibration and instrument signature removal (ISR) data processing pipelines for DP2. Finally, we summarize the known limitations observed in DP2 production and outline expected algorithmic improvements for the first public LSST data release (Data Release~1, DR1).

79 ASTRONOMY AND ASTROPHYSICS↗

Algorithms for Diffractive Microscopy (Final Report)

The phenomenal coherence and brightness of x-ray free-electron laser light sources, such as the LCLS at SLAC, have the potential of revolutionizing the investigation of structure and dynamics in the nano-domain. However, this potential will go unrealized without a similar revolution in the way the data are analyzed. While it is true that the ambitious design parameters of the LCLS have been achieved, the prospects of realizing the most publicized goal of this instrument — the imaging of individual bio-particles — remains daunting. Even with 10 12 photons per x-ray pulse, the feebleness of the scattering process represents a fundamental limit that no amount of engineering ingenuity can overcome. Large bio-molecules will scatter on the order of only 10 3 photons per pulse into a detector with 106 pixels; the diffraction “images” will be virtually indistinguishable from noise. Averaging such noisy signals over many pulses is not possible because the particle orientation cannot be controlled. Each noisy laser snapshot is thus confounded by the unknown viewpoint of the particle. Given the heavy DOE investment in LCLS and the profound technical challenges facing single-particle imaging, the final two years of this project have concentrated on this effort. We are happy to report that we succeeded in developing an extremely efficient algorithm that can reconstruct the shapes of particles at even the extremes of noise expected in future LCLS experiments with single bio-particles. Since this is the most important outcome of this project, the major part of this report documents this accomplishment. The theoretical techniques that were developed for the single-particle imaging project have proved useful in other imaging problems that are described at the end of the report.

77 NANOSCIENCE AND NANOTECHNOLOGY↗

Exploring End-to-end Deep Learning Applications for Event Classification at CMS

An essential part of new physics searches at the Large Hadron Collider (LHC) at CERN involves event classification, or distinguishing potential signal events from those coming from background processes. Current machine learning techniques accomplish this using traditional hand-engineered features like particle 4-momenta, motivated by our understanding of particle decay phenomenology. While such techniques have proven useful for simple decays, they are highly dependent on our ability to model all aspects of the phenomenology and detector response. Meanwhile, powerful deep learning algorithms are capable of not only training on high-level features, but of performing feature extraction. In computer vision, convolutional neural networks have become the state-of-the-art for many applications. Motivated by their success, we apply deep learning algorithms to low-level detector data from the 2012 CMS Simulated Open Data to directly learn useful features, in what we call, end-to-end event classification. We demonstrate the power of this approach in the context of a physics search and offer solutions to some of the inherent challenges, such as image construction, image sparsity, combining multiple sub-detectors, and de-correlating the classifier from the search observable, among others.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Robust Decentralized Learning Using ADMM With Unreliable Agents

Many signal processing and machine learning problems can be formulated as consensus optimization problems which can be solved efficiently via a cooperative multi-agent system. However, the agents in the system can be unreliable due to a variety of reasons: noise, faults and attacks. Providing erroneous updates leads the optimization process in a wrong direction, and degrades the performance of distributed machine learning algorithms. This paper considers the problem of decentralized learning using ADMM in the presence of unreliable agents. First, we rigorously analyze the effect of erroneous updates (in ADMM learning iterations) on the convergence behavior of the multi-agent system. We show that the algorithm linearly converges to a neighborhood of the optimal solution under certain conditions and characterize the neighborhood size analytically. Next, we provide guidelines for network design to achieve a faster convergence to the neighborhood. Here, we also provide conditions on the erroneous updates for exact convergence to the optimal solution. Finally, to mitigate the influence of unreliable agents, we propose ROAD , a robust variant of ADMM, and show its resilience to unreliable agents with an exact convergence to the optimum.

97 MATHEMATICS AND COMPUTING↗

Mining proteomic data to expose protein modifications in Methanosarcina mazei strain Gö1

Proteomic tools identify constituents of complex mixtures, often delivering long lists of identified proteins. The high-throughput methods excel at matching tandem mass spectrometry data to spectra predicted from sequence databases. Unassigned mass spectra are ignored, but could, in principle, provide valuable information on unanticipated modifications and improve protein annotations while consuming limited quantities of material. Strategies to “mine” information from these discards are presented, along with discussion of features that, when present, provide strong support for modifications. In this study we mined LC-MS/MS datasets of proteolytically-digested concanavalin A pull down fractions from Methanosarcina mazei Gö1 cell lysates. Analyses identified 154 proteins. Many of the observed proteins displayed post-translationally modified forms, including O -formylated and methyl-esterified segments that appear biologically relevant (i.e., not artifacts of sample handling). Interesting cleavages and modifications (e.g., S -cyanylation and trimethylation) were observed near catalytic sites of methanogenesis enzymes. Of 31 Methanosarcina protein N -termini recovered by concanavalin A binding or from a previous study, only M. mazei S -layer protein MM1976 and its M. acetivorans C2A orthologue, MA0829, underwent signal peptide excision. Experimental results contrast with predictions from algorithms SignalP 3.0 and Exprot, which were found to over-predict the presence of signal peptides. Proteins MM0002, MM0716, MM1364, and MM1976 were found to be glycosylated, and employing chromatography tailored specifically for glycopeptides will likely reveal more. This study supplements limited, existing experimental datasets of mature archaeal N -termini, including presence or absence of signal peptides, translation initiation sites, and other processing. Methanosarcina surface and membrane proteins are richly modified.

59 BASIC BIOLOGICAL SCIENCES↗

MO-F-CAMPUS-J-05: Verification for Prompt Gamma Ray Imaging During Proton Boron Fusion Therapy

Purpose: The purpose of this study is to verify the acquisition of the three dimensional single photon emission computed tomography (SPECT) image using prompt gamma ray originated from proton boron fusion therapy (PBFT). Methods: The real-time imaging system during the PBFT was simulated to acquire the tomographic image of the prompt gamma ray from treated tumor by the proton boron reaction, using Monte Carlo simulation (MCNPX). We acquired percentage depth dose (PDD) of the proton beam in the water phantom including the boron uptake region (BUR), the energy spectra of the prompt gamma ray and tomographic image which can show treated tumor regions. The prompt gamma ray image was reconstructed using maximum likelihood estimation maximization (MLEM) reconstruction algorithm with 64 projections. In addition, in order to evaluate the reconstructed image, the image profiles between BURs were extracted from the image. Results: In the PDD results, the Bragg-peak was amplified definitely when the proton’s maximum dose level was located at the BUR. This amplification is based on the generation of alpha particles. In addition, the prompt gamma ray peak of 719 keV was observed from the energy spectrum. Through the previous process, the tomographic image of prompt gamma ray from the BUR was reconstructed. The line profile was extracted from image including BURs, and it shows high signal-to-noise ratio. Conclusion: We confirmed that the real-time prompt gamma ray image during the PBFT was successfully deducted, and results of quantitative image analysis show good agreement with the original pattern of the BUR. This study first verified the imaging capability of the PBFT-SPECT system. In conclusion, the PBFT-SPECT system can realize the treated tumor monitoring during the PBFT.

60 APPLIED LIFE SCIENCES↗

Graphical Optimization of Spectral Shift Reconstructions for Optical Backscatter Reflectometry

Optical backscatter reflectometry (OBR) is an interferometric technique that can be used to measure local changes in temperature and mechanical strain based on spectral analyses of backscattered light from a singlemode optical fiber. The technique uses Fourier analyses to resolve spectra resulting from reflections occurring over a discrete region along the fiber. These spectra are cross-correlated with reference spectra to calculate the relative spectral shifts between measurements. The maximum of the cross-correlated spectra—termed quality—is a metric that quantifies the degree of correlation between the two measurements. Recently, this quality metric was incorporated into an adaptive algorithm to (1) selectively vary the reference measurement until the quality exceeds a predefined threshold and (2) calculate incremental spectral shifts that can be summed to determine the spectral shift relative to the initial reference. Using a graphical (network) framework, this effort demonstrated the optimal reconstruction of distributed OBR measurements for all sensing locations using a maximum spanning tree (MST). By allowing the reference to vary as a function of both time and sensing location, the MST and other adaptive algorithms could resolve spectral shifts at some locations, even if others can no longer be resolved.

47 OTHER INSTRUMENTATION↗

A vibro-haptic human-machine interface for structural health monitoring

The structural health monitoring (SHM) community’s goal has been to endow physical systems with a nervous system not unlike those commonly found in living organisms. Typically the SHM community has attempted to do this by instrumenting structures with a variety of sensors, and then applying various signal processing and classification procedures to the data in order to detect the presence of damage, the location of damage, the severity of damage, and to estimate the remaining useful life of the structure. This procedure has had some success, but we are still a long way from achieving the performance of nervous systems found in biology. This is primarily because contemporary classification algorithms do not have the performance required. In many cases expert judgment is superior to automated classification. This work introduces a new paradigm. We propose interfacing the human nervous system to the distributed sensor network located on the structure and developing new techniques to enable human-machine cooperation. Results from the field of sensory substitution suggest this should be possible. This study investigates a vibro-haptic human-machine interface for SHM. The investigation was performed using a surrogate three-story structure. The structure features three nonlinearity-inducing bumpers to simulate damage. Accelerometers are placed on each floor to measure the response of the structure to a harmonic base excitation. The accelerometer measurements are preprocessed. As a result, the preprocessed data is then encoded encoded as a vibro-tactile stimulus. Human subjects were then subjected to the vibro-tactile stimulus and asked to characterize the damage in the structure.

42 ENGINEERING↗

Multi-Source Machine Learning and Thermoplastics Enhanced Aerostructure Manufacturing (mTEAM)

RTX Technology Research Center (RTRC), together with Collins Aerospace (Collins) and Oak Ridge National Laboratory (ORNL) has developed an Artificial Intelligence (AI) / Machine Learning (ML) guided solution to advance the manufacturing and assembly of high performance and lightweight thermoplastic composite (TPC) aerospace products. The solution aims to lower risk, cost and lead time for induction heating based welding and consolidation processes for TPC structure. The cost and lead time of part and material specific process development for induction welding (IW) and induction consolidation will be reduced by replacing traditional empirical methods with optimization methods that merge AI/ML and physics-based process simulations and process experiments with sensing and controls. TPC-IW process development is empirical in nature, and uncertainties in material & process behavior exist near & far from the induction coil. Physics-based simulations can be leveraged directly for process optimization but can be too computationally expensive to run in high fidelity and real time to do robust process optimization. The key impact of successful TPC induction consolidation and welding is cost & lead time reduction for part & material specific consolidation and welding recipes. This is an enabler for more rapid deployment of TPC structures via joining assembly, which can reduce energy & cost intensive usage of autoclaves & ovens. The solution aimed to advance the U.S. Department of Energy’s interests in using thermoplastics and automation in composite manufacturing for improvement of products for existing markets via increased production speeds, reduced costs, and lowered use of energy. Welded TPC structures can offer significant weight & energy savings for high-value commercial aerospace & industrial applications compared to metal & thermoset composite structures assembled by mechanical fastening and/or adhesive bonding. The project was organized into two Budget Periods. Budget Period 1 (BP1) was 15 months and its goal was to perform ML process optimization framework development & deployment on lab-coupon aerostructure components. A Go/No-Go Review was performed at the end of BP1 to verify fulfilment of key tasks & milestones to justify a Go Decision to move into the next Budget Period. Budget Period 2 (BP2) was 12 months and its goal was the deployment of the ML framework for ML process optimization of pilot industrial scale aerostructure components. The overall project aim was to develop & demonstrate ML-enhanced modeling framework that learns process-property mapping from multiple data sources at different fidelities. During BP1, the team accomplished key tasks & milestones to demonstrate the concept of multi-source ML for TPC aerostructure consolidation and assembly. First, the team completed documentation of induction based TPC heating requirements including baseline metrics to compare measured results against. Next the team completed demonstration of data generation from physics-based simulations for ML surrogate model generation and demonstrated the integration of physics-based simulation data into multi-source AI/ML algorithms. In parallel, the team established the lab-coupon scale induction welding system and completed a process to label and reduce generated data from physics-based simulation and experiments for ML surrogate models to enable multi-source ML model training & testing. To complete BP1, the team integrated physics-based simulation data and experimental data into multi-source ML algorithms. This was based on the team completing ML deployment of the induction welding on a lab system at RTRC and AI/ML deployment on existing induction welding line at Collins. ORNL visited both Collins and RTRC sites to witness the TPC induction welding process. Then, ORNL designed and constructed a new version of their vision-based sensing system better adapted to acquire process signals of the TPC induction welding process for process anomaly and defect detection. In BP2, the team accomplished key tasks & milestones to scale up multi-source ML for TPC aerostructure consolidation and assembly from the lab-coupon scale to the pilot-industrial scale. In BP2, the team demonstrated real time anomaly & defect detection via experiments performed by ORNL & RTRC. The team completed ML-optimization heating trials for TPC induction consolidation at Collins, and the team confirmed pilot industrial scale experimental data from Collins was compatible with the developed ML pipeline from RTRC. The team completed sub-element scale ML process optimization demonstration at RTRC, where the team leveraged RTRC’s robotic TPC welding setup to de-risk the ML process optimization by performing ML analysis of recorded temperatures to account for complex part features. Then, the team applied its ML-derived control strategies and ML process optimization framework at Collins to the pilot-industrial scale on a demo skin-stiffener part representative of a nacelle aerostructure fan cowl section. The key innovation is the AI/ML framework enabling effective process development of high performance, lightweight, energy efficient TPCs for composite aircraft structures.

36 MATERIALS SCIENCE↗

TU-AB-201-11: A Novel Theoretical Framework for MRI-Only Image Guided LDR Prostate and Breast Brachytherapy Implant Dosimetry

Purpose: To propose a novel framework for accurate model-based dose calculations using only MR images for LDR prostate and breast seed implant brachytherapy. Methods: Model-based dose calculation methodologies recommended by TG-186 require further knowledge about specific tissue composition, which is challenging with MRI. However, relying on MRI-only for implant dosimetry would reduce the soft tissue delineation uncertainty, costs, and uncertainties associated with multi-modality registration and fusion processes. We propose a novel framework to address this problem using quantitative MRI acquisitions and reconstruction techniques. The framework includes three steps: (1) Identify the locations of seeds(2) Identify the presence (or absence) of calcification(s)(3) Quantify the water and fat content in the underlying tissueSteps (1) and (2) consider the sources that limit patient dosimetry, particularly the inter-seed attenuation and the calcified regions; while step (3) targets the quantification of the tissue composition to consider the heterogeneities in the medium. Our preliminary work has shown that the seeds and the calcifications can be identified with MRI using both the magnitude and the phase images. By employing susceptibility-weighted imaging with specific post-processing techniques, the phase images can be further explored to distinguish the seeds from the calcifications. Absolute quantification of tissue, water, and fat content is feasible and was previously demonstrated in phantoms and in-vivo applications, particularly for brain diseases. The approach relies on the proportionality of the MR signal to the number of protons in an image volume. By employing appropriate correction algorithms for T1 - and T2*-related biases, B1 transmit and receive field inhomogeneities, absolute water/fat content can be determined. Results: By considering calcification and interseed attenuation, and through the knowledge of water and fat mass density, accurate patient-specific implant dosimetry can be achieved with MRI-only. Conclusion: The proposed framework showed that model-based dose calculation is feasible using MRI-only state-of-the-art techniques.

60 APPLIED LIFE SCIENCES↗

Fully Distributed Multi-Material Magnetic Sensing Structures for Multiparameter DAS Applications

This dissertation demonstrates the first of its kind distributed magnetic field sensor based on a fiber optic distributed acoustic sensing (DAS) scheme. Ferromagnetic nickel and Metglas® were dispersed internally within a fiber optic preform and then drawn on an in-house fiber optic draw tower to lengths in the kilometers. Due to the close proximity of the ferromagnetic metals and fiber optic core, the magnetostrictive strain response of the ferromagnetic materials when exposed to a magnetic field would perturbate within the fiber cladding and transfer that strain, internally, to the fiber optic core. Strain resulting from the magnetostrictive effect allows the DAS based sensor to accurately translate strain into readable magnetic field data. Due to the high sensitivity seen in this sensor design, multiparameter sources, acoustic and magnetic fields, were tested and validated and a three dimensional magnetic-field vector sensor was proposed. Numerical analysis of the novel sensor design was first implemented using COMSOL Multiphysics, where inputs such as magnetostrictive element shape, size, distance, and number were first investigated. Upon optimizing system constraints, the sensor design was further modified such that single mode operation was consistent across multiple fiber draws while retaining high strain transfer from the ferromagnetic elements to the fiber optic core. Ferromagnetic material selection was evaluated as a function of the saturation magnetostriction constants and a total of 4 modules were used to fully characterize the complex physics involved in this sensor design. All fabrication and testing were performed in-house using a full scale 3-story fiber draw tower and custom environmental testing stations to imitate naturally occurring events such as magnetic or acoustic point sources. A unique stacking method was used to embed ferromagnetic nickel and Metglas® into a fiber optic preform which when combined with a custom fiber draw process resulted in consistent multi-material fibers drawn to lengths of 1-km. In-house testing facilities included different types of electromagnetic generators, in addition to a soil test bed, and an outdoor test bed which allowed 100 meters of fiber to be tested simultaneously. All tested sensors demonstrated high strain transfer capabilities on the order of 0.01-10 μϵ depending on the materials used, ferromagnetic rod number, and core to metal spacing. Due to the sensitivity of the system the difference between AC and DC was distinct, and directional magnetostriction was studied. Transverse and longitudinal magnetic wave propagation was controlled through a solenoid and rectangular Helmholtz coil, both built in-house. A three-dimensional magnetic field vector sensor was proposed due to the success of the magnetic field sensor, and a design was proposed and initially tested to validate direction as a function of field strength and distance. To summarize, this dissertation explores the first fully distributed magnetic field sensor using DAS based techniques and one of the first multi-material fiber draw processes which can produce consistent single mode fiber up to 1-km. Due to extensive FEA modeling, multiple iterations of the magnetic sensor were fully characterized and an equation describing the relationship between sensor design and strain transfer has been created and validated experimentally. Multi-parameter tests including acoustic and magnetic fields were implemented and an algorithm was developed to separate the mixed signals. Finally, a test was performed to demonstrate the feasibility of sensing magnetic fields directionally. Cumulative results demonstrate a high-quality sensor alternative to current designs which may surpass other magnetic sensors due to innate multi-parameter capabilities, in addition to the inexpensive production cost and extremely long operating lengths.

02 PETROLEUM↗

Smoothing of respiratory motion traces for motion-compensated radiotherapy

Purpose: The CyberKnife system has been used successfully for several years to radiosurgically treat tumors without the need for stereotactic fixation or sedation of the patient. It has been shown that tumor motion in the lung, liver, and pancreas can be tracked with acceptable accuracy and repeatability. However, highly precise targeting for tumors in the lower abdomen, especially for tumors which exhibit strong motion, remains problematic. Reasons for this are manifold, like the slow tracking system operating at 26.5 Hz, and using the signal from the tracking camera ''as is''. Since the motion recorded with the camera is used to compensate for system latency by prediction and the predicted signal is subsequently used to infer the tumor position from a correlation model based on x-ray imaging of gold fiducials around the tumor, camera noise directly influences the targeting accuracy. The goal of this work is to establish the suitability of a new smoothing method for respiratory motion traces used in motion-compensated radiotherapy. The authors endeavor to show that better prediction--With a lower rms error of the predicted signal--and/or smoother prediction is possible using this method. Methods: The authors evaluated six commercially available tracking systems (NDI Aurora, PolarisClassic, Polaris Vicra, MicronTracker2 H40, FP5000, and accuTrack compact). The authors first tracked markers both stationary and while in motion to establish the systems' noise characteristics. Then the authors applied a smoothing method based on the a trous wavelet decomposition to reduce the devices' noise level. Additionally, the smoothed signal of the moving target and a motion trace from actual human respiratory motion were subjected to prediction using the MULIN and the nLMS{sub 2} algorithms. Results: The authors established that the noise distribution for a static target is Gaussian and that when the probe is moved such as to mimic human respiration, it remains Gaussian with the exception of the FP5000 and the Aurora systems. The authors also showed that the proposed smoothing method can indeed be used to filter noise. The signal's jitter dropped by as much as 95% depending on the tracking system employed. Subsequently, the 3D prediction error (rms) for a prediction horizon of 150 ms on a synthetic signal dropped by up to 37% when using a normalized LMS prediction algorithm (nLMS{sub 2}) and hardly changed when using a MULIN algorithm. When smoothing a real signal obtained in our laboratory, the improvement of prediction was similar: Up to 30% for both the nLMS{sub 2} and the best MULIN algorithm. The authors also found a noticeable increase in smoothness of the predicted signal, the relative jitter dropped by up to 95% on the real signal, and on the simulated signal. Conclusions: In conclusion, the authors can say that preprocessing of marker data is very useful in motion-compensated radiotherapy since the quality of prediction increases. This will result in better performance of the correlation model. As a side effect, since the prediction of a preprocessed signal is also less noisy, the authors expect less robot vibration resulting in better targeting accuracy and less strain on the robot gears.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Implementing Neural Adaptive Filtering in Engineered Detection Systems

The retina plays an important role in animal vision --- namely to pre-process visual information before sending it to the brain. The goal of this LDRD was to develop models of motion-sensitive retinal cells for the purpose of developing retinal-inspired algorithms to be applied to real-world data specific to Sandia's national security missions. We specifically focus on detection of small, dim moving targets amidst varying types of clutter or distractor signals. We compare a classic motion-sensitive model, the Hassenstein-Reichardt model, to a model of the OMS (object motion- sensitive) cell, and find that the Reichardt model performs better under continuous clutter (e.g. white noise) but is very sensitive to particular stimulus conditions (e.g. target velocity). We also demonstrate that lateral inhibition, a ubiquitous characteristic of neural circuitry, can effect target-size tuning, improving detection specifically of small targets.

47 OTHER INSTRUMENTATION↗