Search NASA⌕ Search

SEARCH · Search NASA

Results for “Algorithm Development and Nuclear Data”

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 703 records · Page 39

FY25 Progress Report: SRNL Analysis of ICCWR LCM and WAMS data for Corrosion and Cracking

Algorithms for Machine Learning (ML) and image analysis for the 3013 Surveillance Program have been developed in an ongoing collaborative effort by the Savannah River National Laboratory (SRNL) and the University of South Carolina (USC). The objective of the algorithms is to automate the identification of corrosion and cracks in the Inner Container Closure Weld Region (ICCWR) of the canister system used to store Pu-bearing material. Data for corrosion and cracking is collected from large binary files generated by a Laser Confocal Microscope (LCM), the Wide Area 3D Measurement System (WAMS), or, in a recent proposal, by a Scanning Electron Microscope (SEM). The ML software uses the physical attributes in the data files (e.g., one or more of: height, color, and 16-bit grayscale values as functions of position in a plane projection) to detect signs of surface corrosion and cracking after being trained on similar data with the features to be detected. Although the initial scope included screening for broader indicators of corrosion, e.g., pitting, the identification of potential cracks was prioritized for the past several years at the request of program leadership.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Gaussian mixture model clustering algorithms for the analysis of high-precision mass measurements

The development of the phase-imaging ion-cyclotron resonance (PI-ICR) technique for use in Penning trap mass spectrometry (PTMS) increased the speed and precision with which PTMS experiments can be carried out. In PI-ICR, data sets of the locations of individual ion hits on a detector are created showing how ions cluster together into spots according to their cyclotron frequency. Ideal data sets would consist of a single, 2D-spherical spot with no other noise, but in practice data sets typically contain multiple spots, non-spherical spots, or significant noise, all of which can make determining the locations of spot centers non-trivial. A method for assigning groups of ions to their respective spots and determining the spot centers is therefore essential for further improving precision and confidence in PI-ICR experiments. Here, we present the class of Gaussian mixture model (GMM) clustering algorithms as an optimal solution. We show that on simulated PI-ICR data, several types of GMM clustering algorithms perform better than other clustering algorithms over a variety of typical scenarios encountered in PI-ICR. The mass spectra of 163Gd, 163 mGd, 162 Tb, and 162 mTb measured using PI-ICR at the Canadian Penning trap mass spectrometer were checked using GMMs, producing results that were in close agreement with the previously published values.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Automated UF 6 Cylinder Enrichment Assay: Status of the Hybrid Enrichment Verification Array (HEVA) Project: POTAS Phase II

Pacific Northwest National Laboratory (PNNL) intends to automate the UF 6 cylinder nondestructive assay (NDA) verification currently performed by the International Atomic Energy Agency (IAEA) at enrichment plants. PNNL is proposing the installation of a portal monitor at a key measurement point to positively identify each cylinder, measure its mass and enrichment, store the data along with operator inputs in a secure database, and maintain continuity of knowledge on measured cylinders until inspector arrival. This report summarizes the status of the research and development of an enrichment assay methodology supporting the cylinder verification concept. The enrichment assay approach exploits a hybrid of two passively-detected ionizing-radiation signatures: the traditional enrichment meter signature (186-keV photon peak area) and a non-traditional signature, manifested in the high-energy (3 to 8 MeV) gamma-ray continuum, generated by neutron emission from UF 6 . PNNL has designed, fabricated, and field-tested several prototype assay sensor packages in an effort to demonstrate proof-of-principle for the hybrid assay approach, quantify the expected assay precision for various categories of cylinder contents, and assess the potential for unsupervised deployment of the technology in a portal-monitor form factor. We refer to recent sensor-package prototypes as the Hybrid Enrichment Verification Array (HEVA). The report provides an overview of the assay signatures and summarizes the results of several HEVA field measurement campaigns on populations of Type 30B UF 6 cylinders containing low-enriched uranium (LEU), natural uranium (NU), and depleted uranium (DU). Approaches to performance optimization of the assay technique via radiation transport modeling are briefly described, as are spectroscopic and data-analysis algorithms.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

2D lag and signal nonlinearity correction in an amorphous silicon EPID and their impact on pretreatment dosimetric verification

Purpose: This investigation provides measurements of signal lag and nonlinearity separately for the Varian aS500 electronic portal imaging device (EPID), and an algorithm to correct for these effects in 2D; their potential impact on intensity modulated radiation therapy (IMRT) verification is also investigated. The authors quantify lag, as a function of both delivered monitor units (MU) and time, by using a range of MUs delivered at a clinically used rate of 400 MU/min. Explicit cumulative lag curves are thus determined for a range of MUs and times between the end of irradiation and the end of image acquisition. Signal nonlinearity is also investigated as a function of total MUs delivered. The family of cumulative lag curves and signal nonlinearity are then used to determine their effects on dynamic multileaf collimator (MLC) (IMRT) deliveries, and to correct for theses effects in 2D. Methods: Images acquired with an aS500 EPID and Varis Portal-Vision software were used to quantify detector lag and signal-nonlinearity. For the signal lag investigation, Portal-Vision's service monitor was used to acquire EPID images at a rate of 8 frames/s. The images were acquired during irradiation and 66 s thereafter, by inhibiting the M-holdoff-In signal of the Linac for a range of 4.5-198.5 MUs. Relative cumulative lag was calculated by integrating the EPID signal for a time after beam-off, and normalizing this to the integrated EPID signal accumulated during radiation. Signal nonlinearity was studied by acquiring 10 Multiplication-Sign 10 cm{sup 2} open-field EPID images in 'integrated image' mode for a range of 2-500 MUs, and normalized to the 100 MU case. All data were incorporated into in-house written software to create a 2D correction map for these effects, using the field's MLC file and a field-specific calculated 2D 'time-map,' which keeps track of the time elapsed from the last fluence delivered at each given point in the image to the end of the beam delivery. Results: Relative cumulative lag curves reveal that the lag alone can deviate the EPID's perceived dose by as large as 6% (1 MU delivery, 60 s postirradiation). For signal nonlinearity relative to 100 MU, EPID signals per MU of 0.84 and 1.01 were observed for 2 and 500 MUs, respectively. Correction maps were applied to a 1 cm sweeping-window 14 Multiplication-Sign 14 cm{sup 2} field and clinical head-and-neck IMRT field. A mean correction of 1.028 was implemented in the head-and-neck field, which significantly reduced lag-related asymmetries in the EPID images, and restored linearity to the EPID imager's dose response. Corrections made to the sweeping-field showed good agreement with the treatment planning system-predicted field, yielding an average percent difference of 0.05%{+-} 0.91%, compared to the -1.32%{+-} 1.02% before corrections, or 1.75%{+-} 1.04% when only a signal nonlinearity correction is made. Conclusions: Lag and signal-nonlinearity have been quantified for an aS500 EPID imager, and an effective 2D correction method has been developed which effectively removes nonlinearity and lag effects. Both of these effects were shown to negatively impact IMRT verifications. Especially fields that involve prolonged irradiation and small overall MUs should be corrected for in 2D.

60 APPLIED LIFE SCIENCES↗

A reference-area-free strain mapping method using precession electron diffraction data

Here, in this work, we developed a method using precession electron diffraction data to map the residual elastic strain at the nano-scale. The diffraction pattern of each pixel was first collected and denoised. Template matching was then applied using the center spot as the mask to identify the positions of the diffraction disks. Statistics of distances between the selected diffracted disks enable the user to make an informed decision on the reference and to generate strain maps. Strain mapping on an unstrained single crystal sapphire shows the standard deviation of strain measurement is 0.5%. With this method, we were able to successfully measure and map the residual elastic strain in VO 2 on sapphire and martensite in a Ni 50.3 Ti 29.7 Hf 20 shape memory alloy. This approach does not require the user to select a “strain-free area” as a reference and can work on datasets even with the crystals oriented away from zone axes. This method is expected to provide a robust and more accessible alternative means of studying the residual strain of various material systems that complements the existing algorithms for strain mapping.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Nuclear Fission Investigation with Twin Ionization Chamber

The purpose of the present paper was to report the recent results, obtained in development of digital pulse processing mathematics for prompt fission neutron (PFN) investigation using twin ionization chamber (TIC) along with fast neutron time-of-flight detector (ND). Due to well known ambiguities in literature (see refs. [4, 6, 9 and 11]), concerning a pulse induction on TIC electrodes by FF ionization, we first presented detailed mathematical analysis of fission fragment (FF) signal formation on TIC anode. The analysis was done using Ramo-Shockley theorem, which gives relation between charged particle motion between TIC electrodes and so called weighting potential. Weighting potential was calculated by direct numerical solution of Laplace equation (neglecting space charge) for the TIC geometry and ionization, caused by FF. Formulae for grid inefficiency (GI) correction and digital pulse processing algorithms for PFN time-of-flight measurements and pulse shape analysis are presented and discussed.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Materials Characterization, Prediction, and Control Project: Characterization of 316L Stainless Steel after Solid Phase Processing using Ultrasonic NDE Method

The Pacific Northwest National Laboratory undertook the Materials Characterization, Prediction, and Control Laboratory Directed Research and Development Project to advance understanding of nuclear material processing and enable multifold acceleration in the development and qualification of new material systems produced via advanced manufacturing methods, such as solid phase processing, for use in national security and advanced energy applications (Smith 2021). A motivation of the Materials Characterization, Prediction, and Control Project was to demonstrate ultrasonic testing as a nondestructive evaluation method to complement traditional destructive methods for characterizing material microstructure with emphasis on grain size determination using a method that may have future applications for real-time inline process monitoring. The objective of the work described in this report is to establish the process and an analysis method for measuring grain sizes of polycrystalline metals with ultrafine grains using ultrasonic shear wave backscattering, building on prior studies on coarser-grained material. The work involves five tasks: Measured ultrasonic backscattering experimentally for a series of 316L stainless steel specimens with various grain sizes made by friction stir processing. Calculated ultrasonic backscattering coefficients from experimental data based on a physical measurement model. Measured ground truth grain sizes of the specimens from electron backscatter diffraction grain boundary images using a generalization of the ASTM E112 (ASTM 2021) intercept method. Built a curve of ultrasonic backscattering coefficients versus the ground truth intercept-based grain sizes to determine the correlation between mean grain sizes and ultrasonic measurements. Demonstrated the ability of using the correlation curve to deduce grain sizes with measured ultrasonic backscattering coefficients for a few 316L stainless steel specimens whose grain sizes were unknown beforehand but were targeted to be an extrapolation to larger grain sizes than used to formulate the correlation curves. Experimental procedures and computational algorithms are developed and validated for these tasks. This work establishes an ultrasonic technique for characterizing material microstructure with ultrafine grains that are often resulted by solid-phase processing. The technique is nondestructive, and it has the potential to be used for real time inline process monitoring. This work successfully demonstrates the viability of an ultrasonic nondestructive evaluation method for microstructural characterization of material having ultrafine grain structure (as small as 1?mm) and produced by an advanced manufacturing method. This includes a demonstration of the method to extrapolate to other conditions. While not demonstrated here, the method is expected to be viable for in-line, or near-inline, process monitoring in advanced manufacturing applications with suitable consideration for access of instrumentation to the material being manufactured.

316 L Stainless Steel↗

TU-A-12A-12: Improved Airway Measurement Accuracy for Low Dose Quantitative CT (qCT) Using Statistical (ASIR), at Reduced DFOV, and High Resolution Kernels in a Phantom and Swine Model

Purpose: Quantitative accuracy of Iodine Hounsfield Unit (HU) in conventional single-kVp scanning is susceptible to beam-hardening effect. Dual-energy CT has unique capabilities of quantification using monochromatic CT images, but this scanning mode requires the availability of the state-of-the-art CT scanner and, therefore, is limited in routine clinical practice. Purpose of this work was to develop a beam-hardening-correction (BHC) for single-kVp CT that can linearize Iodine projections at any nominal energy, apply this approach to study Iodine response with respect to keV, and compare with dual-energy based monochromatic images obtained from material-decomposition using 80kVp and 140kVp. Methods: Tissue characterization phantoms (Gammex Inc.), containing solid-Iodine inserts of different concentrations, were scanned using GE multi-slice CT scanner at 80, 100, 120, and 140 kVp. A model-based BHC algorithm was developed where Iodine was estimated using re-projection of image volume and corrected through an iterative process. In the correction, the re-projected Iodine was linearized using a polynomial mapping between monochromatic path-lengths at various nominal energies (40 to 140 keV) and physically modeled polychromatic path-lengths. The beam-hardening-corrected 80kVp and 140kVp images (linearized approximately at effective energy of the beam) were used for dual-energy material-decomposition in Water-Iodine basis-pair followed by generation of monochromatic images. Characterization of Iodine HU and noise in the images obtained from singlekVp with BHC at various nominal keV, and corresponding dual-energy monochromatic images, was carried out. Results: Iodine HU vs. keV response from single-kVp with BHC and dual-energy monochromatic images were found to be very similar, indicating that single-kVp data may be used to create material specific monochromatic equivalent using modelbased projection linearization. Conclusion: This approach may enable quantification of Iodine contrast enhancement and potential reduction in injected contrast without using dual-energy scanning. However, in general, dual-energy scanning has unique value in material characterization and quantification, and its value cannot be discounted. GE Healthcare Employee.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

SU-E-T-637: Proton Aperture Quality Assurance Using Computed Tomography

Purpose: To develop an automated algorithm for aperture quality assurance for double scattering proton radiotherapy. Methods: Anterior-Posterior scout images were acquired on a CT scanner for five brass apertures (2 large, 3 small) with 120 kVp and 10mA. A semi-automated algorithm was developed for the QA process. The input includes a user selecting the scout image and the associated patient plan, field and aperture size. The program automatically thresholds the scout image. Next, a connectivity algorithm is used to determine the points associated with the central shape to obtain the contour. To compare with the plan contour, the program converts the detected and plan contours into polar coordinates, interpolates the data to a 1 degree spaced grid, and determines the differences in radial distance at each grid point. Results: The mean and maximum difference and the percentage of points with distance differences less than 1.25 mm (due to the divergent cut of the aperture) between the detected and plan field aperture contours were obtained. A repeatability coefficient was derived based on repeated scanning and processing of three of the apertures. The mean difference for five apertures was 0.44 +/−0.08 mm. The maximum difference in distance was 1.2 +/−0.23 mm for all apertures. The Repeatability Coefficient was +/−0.038 mm, indicating the technique is highly repeatable. The mean percent of points with distance less than 1.25 mm was 97.41 +/−0.35%. Conclusion: Automated CT scout image based proton aperture QA is feasible, saves time, and provides a quantitative metric for proton patient specific aperture QA.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Accurate Automatic Delineation of Heterogeneous Functional Volumes in Positron Emission Tomography for Oncology Applications

Purpose: Accurate contouring of positron emission tomography (PET) functional volumes is now considered crucial in image-guided radiotherapy and other oncology applications because the use of functional imaging allows for biological target definition. In addition, the definition of variable uptake regions within the tumor itself may facilitate dose painting for dosimetry optimization. Methods and Materials: Current state-of-the-art algorithms for functional volume segmentation use adaptive thresholding. We developed an approach called fuzzy locally adaptive Bayesian (FLAB), validated on homogeneous objects, and then improved it by allowing the use of up to three tumor classes for the delineation of inhomogeneous tumors (3-FLAB). Simulated and real tumors with histology data containing homogeneous and heterogeneous activity distributions were used to assess the algorithm's accuracy. Results: The new 3-FLAB algorithm is able to extract the overall tumor from the background tissues and delineate variable uptake regions within the tumors, with higher accuracy and robustness compared with adaptive threshold (T{sub bckg}) and fuzzy C-means (FCM). 3-FLAB performed with a mean classification error of less than 9% +- 8% on the simulated tumors, whereas binary-only implementation led to errors of 15% +- 11%. T{sub bckg} and FCM led to mean errors of 20% +- 12% and 17% +- 14%, respectively. 3-FLAB also led to more robust estimation of the maximum diameters of tumors with histology measurements, with <6% standard deviation, whereas binary FLAB, T{sub bckg} and FCM lead to 10%, 12%, and 13%, respectively. Conclusion: These encouraging results warrant further investigation in future studies that will investigate the impact of 3-FLAB in radiotherapy treatment planning, diagnosis, and therapy response evaluation.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Evaluation of dose recalculation vs dose deformation in a commercial platform for deformable image registration with a computational phantom

Deformable image registration (DIR) is an important tool for mapping both dose and contours of a new set of planning images when recurrent, that is, adaptive radiotherapy, or further treatment, that is, re-treatment, is required. The aim of this study was to evaluate the need for plan recalculation in deformed anatomies and to develop a reliable workflow for validating the DIR algorithm to be applied to dose warping such as dose accumulation (DA) for adaptive radiation therapy, and dose summation (DS) in the case of re-treatment. A set of 3 computational phantoms was developed to validate the application of B-Spline-based registrations for dose mapping among the various computed tomography image data sets. Two different versions were defined for each phantom to simulate clinical needs: adaptive radiotherapy and re-treatment; a DIR was performed to obtain a displacement vector field (DVF) for dose applications. Comparison of calculated and deformed doses was carried out by means of known markers inside the virtual phantoms. The differences were evaluated using a 3% dose index as acceptance criteria. A paired Wilcoxon signed-rank test was carried out to test the statistical significance of differences within the markers. Significant differences were only observed for the deformed dose in the DA test; no significant differences were observed for recalculated dose values in the DA and DS tests. The dose index is in accordance with these results. The warping dose process obtained by registering various sets of images was validated; a recalculation approach seems to be more accurate for DA purposes and image-guided adaptive radiation therapy (IGART) applications.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

High speed, multi-channel, user programmable digital data acquisition system.

As applications for radiation detection become more demanding, and in turn improvements are made in the technology of radiation detection, there is a need for high speed digital detector readout electronics matching these improvements. Specifically, full control over the on-line processing resources of modern digital electronics is desirable so that researchers can develop custom algorithms for special applications.In the proposed effort, the 500 MHz digital readout electronics previously developed by our company will be redesigned to allow user access to the on-line processing resources. In Phase I, the division of online processing into vendor and user firmware sections has been studied on existing hardware. In Phase II, the hardware will be upgraded to better facilitate the division, and the firmware will be restructured into a robust vendor logic block (providing standard functions such as host I/O, on-board memory I/O, energy computation, MCA spectra, timestamps, waveform capture, run statistics, and triggering and timing) and a user logic block for custom algorithms (with templates and examples for frequently used functions). Investigating several options to divide online processing, it was determined that the most promising approach is to “partition” a single FPGA integrated circuit into a vendor and user section, which is supported in newer devices. The analog front end of the existing electronics proved suitable for most applications, in particular high rate measurements with germanium detectors. The design architecture for new electronics was developed, combining one of the new FPGA device with the analog front end.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

MCNP 6.2.0 Delayed-Particle Production Improvements

Development began on a delayed-particle physics package for the Monte Carlo Radiation transport code MCNPX in 2004 incorporating the CINDER90 depletion code and delayed particle libraries. When MCNP5 was merged with MCNPX to form MCNP6 in 2010, this delayed-particle capability was passed on and has been continuously improved in subsequent versions of MCNP6. Today, the delayed-particle feature in MCNP6 allows for the production of delayed neutrons, gammas, betas, and alpha particles. Future development includes the addition of delayed-positron production in the next release of MCNP6. Delayed particles can be sampled with MCNP using the ACT card in conjunction with the FISSION and NONFISS keywords. The user can specify one or more of the four available delayed-particle types mentioned above on one or both of the FISSION and NONFISS keywords. If a delayed-particle type is selected on the FISSION keyword, then delayed production of that particle from fission products and their progeny will be produced. Likewise, if a delayed-particle type is selected on the NONFISS keyword then delayed production of that particle from non-fission interactions and spontaneous decay will be produced. In some cases, the sheer number of delayed particles that can be produced from nuclear reactions and spontaneous decay (i.e. hundreds of delayed gamma lines when spontaneous fission is possible) can make it difficult to achieve converged solutions in an efficient and timely manner. Truncation methods are the simplest form of variance reduction and can be used to decrease convergence times by limiting calculations to the phase space of interest. This paper introduces several such truncation methods that allow the user to restrict the number of possible delayed particles available for sampling to only those of interest in the problem. It is important to stress that the use of any truncation method should be verified by first assessing the impact of the truncated particles (i.e. by disabling the truncation) for one or more baseline cases. In addition to these new methods, a new delayed-gamma sampling algorithm implemented in MCNP 6.2.0 is introduced. Comparisons to ENDF data and analytical benchmarking with the new algorithm show a substantial improvement in accuracy over previous versions. New improvements to the delayed-particle capability have been added to MCNP 6.2.0. These improvements include an updated treatment of the THRESH keyword which omits delayed-particle lines below a set amplitude, and addition of the PECUT and HLCUT keywords which allow the user to truncate delayed-particles lines based on energy and half-life, respectively. The delayed-gamma sampling algorithm has also been improved to allow exact energy sampling. Comparisons of MCNP calculations directly to ENDF data and analytical benchmarks have verified the significant improvement in the new delayed gamma sampling algorithm. (authors)

97 MATHEMATICS AND COMPUTING↗

SU-C-18A-02: Image-Based Camera Tracking: Towards Registration of Endoscopic Video to CT

Purpose: Endoscopic examinations are routinely performed on head and neck and esophageal cancer patients. However, these images are underutilized for radiation therapy because there is currently no way to register them to a CT of the patient. The purpose of this work is to develop a method to track the motion of an endoscope within a structure using images from standard clinical equipment. This method will be incorporated into a broader endoscopy/CT registration framework. Methods: We developed a software algorithm to track the motion of an endoscope within an arbitrary structure. We computed frame-to-frame rotation and translation of the camera by tracking surface points across the video sequence and utilizing two-camera epipolar geometry. The resulting 3D camera path was used to recover the surrounding structure via triangulation methods. We tested this algorithm on a rigid cylindrical phantom with a pattern spray-painted on the inside. We did not constrain the motion of the endoscope while recording, and we did not constrain our measurements using the known structure of the phantom. Results: Our software algorithm can successfully track the general motion of the endoscope as it moves through the phantom. However, our preliminary data do not show a high degree of accuracy in the triangulation of 3D point locations. More rigorous data will be presented at the annual meeting. Conclusion: Image-based camera tracking is a promising method for endoscopy/CT image registration, and it requires only standard clinical equipment. It is one of two major components needed to achieve endoscopy/CT registration, the second of which is tying the camera path to absolute patient geometry. In addition to this second component, future work will focus on validating our camera tracking algorithm in the presence of clinical imaging features such as patient motion, erratic camera motion, and dynamic scene illumination.

60 APPLIED LIFE SCIENCES↗

A clinically feasible method for the detection of potential collision in proton therapy

Purpose: Potential collision between the patient/couch and the gantry could delay the start of the treatment and reduce clinical efficiency. The ability to accurately detect possible collisions during the treatment planning phase is desired. Such collision detection should account for the specific proton gantry design, the treatment beam configuration, couch orientation, and the patient specific geometry. In this paper the authors developed an approach to detect possible patient-machine collisions using patient treatment plan data. Methods: The geometry of the machine and the patient was reconstructed relative to the isocenter of the proton treatment room. The surface contour of the gantry was first captured from the proton computer aided design and reconstructed to account for specific gantry rotation, snout position, collimator rotation, and range compensator dimensions based on the patient treatment plan data. The patient body and couch contours were captured from the patient's CT DICOM structure file. They were reconstructed relative to the isocenter taking into account treatment couch rotation. For potential collision that occurs at body portions where no CT images exist, scout images are used to construct the body contour. A software program was developed using a ray casting algorithm that was applied to detect collisions by determining if any of the patient and couch contour points fall into the spatial polygons formed by the proton gantry surfaces. Results: Twenty-four patient plans with or without potential collisions were retrospectively identified and analyzed using the collision detection software. In addition, five collision cases were artificially generated using an anthropomorphic phantom. The program successfully detected the collisions in all cases. The calculation time for each case was within 20 s. The software program was implemented in the authors' clinic to detect patient-gantry or gantry-couch collisions in the treatment planning phase. Conclusions: The authors developed a fast and clinically feasible patient-specific collision detection program for proton therapy based on a ray casting algorithm. If incorporated during the treatment planning phase it may lead to improved clinical efficiency. This methodology could also be applied to patient collision detection in photon therapy.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Proposing New RadLex Terms by Analyzing Free-Text Mammography Reports

After years of development, the RadLex terminology contains a large set of controlled terms for the radiology domain, but gaps still exist. We developed a data-driven approach to discover new terms for RadLex by mining a large corpus of radiology reports using natural language processing (NLP) methods. Our system, developed for mammography, discovers new candidate terms by analyzing noun phrases in free-text reports to extend the mammography part of RadLex. Our NLP system extracts noun phrases from free-text mammography reports and classifies these noun phrases as “Has Candidate RadLex Term” or “Does Not Have Candidate RadLex Term.” We tested the performance of our algorithm using 100 free-text mammography reports. An expert radiologist determined the true positive and true negative RadLex candidate terms. We calculated precision/positive predictive value and recall/sensitivity metrics to judge the system’s performance. Finally, to identify new candidate terms for enhancing RadLex, we applied our NLP method to 270,540 free-text mammography reports obtained from three academic institutions. Our method demonstrated precision/positive predictive value of 0.77 (159/206 terms) and a recall/sensitivity of 0.94 (159/170 terms). The overall accuracy of the system is 0.80 (235/293 terms). When we ran our system on the set of 270,540 reports, it found 31,800 unique noun phrases that are potential candidates for RadLex. Our data-driven approach to mining radiology reports can identify new candidate terms for expanding the breast imaging lexicon portion of RadLex and may be a useful approach for discovering new candidate terms from other radiology domains.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

FY21 Progress Report: SRNL Analysis of ICCWR LCM and WAMS data for Corrosion and Cracking

The development of algorithms for machine learning and data analysis for the 3013 Surveillance Program is a collaborative effort by the Savannah River National Laboratory (SRNL) and the University of South Carolina (USC). For corrosion detection, Laser Confocal Microscope (LCM) or Wide Area 3D Measurement System (WAMS) data is extracted from large binary files, with software written to convert the data to physical attributes (e.g., height, color and grayscale values; all as functions of a location in a plane projection). A user-friendly Matlab Graphical User Interface (GUI) that reads data from either LCM or WAMS files was developed to integrate input data with software developed for processing and evaluation. The GUI can selectively download binary data, interrogate data attributes, label data, flag significant features, execute Machine Learning (ML) algorithms, output parameters for trained ML algorithms, report ML model accuracy with respect to labeled data, and generate graphical representations for various analyses. Features can be called out by user-specified thresholds, manual labeling or machine learning algorithms when they have been completed. The ability to rapidly label data is important because of the volume of data required for training machine learning algorithms. The GUI has the flexibility to allow addition of improved ML algorithms, methods for data visualization, and statistical computations. Statistical analyses via the GUI include areas of pits within a defined range of pit depths, correlations between Red-Green-Blue (RGB) or grayscale intensity and relative surface height, covariances between values associated with features, and feature histograms. The development of supervised machine learning algorithms, however, has been hindered by a lack of training data. The machine learning algorithms for crack identification are being refined but require improvements to the true positive rate for crack detection. This shortcoming is an artifact of the limited training data currently available, perhaps more so than the structure of the neural networks. At present, the best results are had from a consensus over an ensemble of randomly generated Deep Neural Network (DNN) or Convolutional Neural Network (CNN) algorithms. Although the consensus accuracy method has yielded optimum true positive and true negative rates in excess of 80%, additional validation testing is necessary. In addition to the suite of LCM data that was initially used, and which represents the majority of the work presented in this report, WAMS image data was also reviewed at a preliminary level. The review included a comparison between image resolution and dynamic range for each method. WAMS (ZON file) image data was found to have a pixel pitch of 3.69μm compared to 1 μm for the LCM (vk4 file) data, which implies a lower resolution for the WAMS images. Conversely, the ratio of dynamic range of the WAMS data to the LCM data was approximately 41:20 for height data, suggesting that information from WAMS should more accurately determine the depth of pits. At present, the significance of the greater dynamic range of the WAMS data relative to the LCM data has not yet been evaluated.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Pu239 Cross-Section Variations Based on Experimental Uncertainties and Covariances

Algorithms and software have been developed for producing variations in plutonium-239 neutron cross sections based on experimental uncertainties and covariances. The varied cross-section sets may be produced as random samples from the multi-variate normal distribution defined by an experimental mean vector and covariance matrix, or they may be produced as Latin-Hypercube/Orthogonal-Array samples (based on the same means and covariances) for use in parametrized studies. The variations obey two classes of constraints that are obligatory for cross-section sets and which put related constraints on the mean vector and covariance matrix that detemine the sampling. Because the experimental means and covariances do not obey some of these constraints to sufficient precision, imposing the constraints requires modifying the experimental mean vector and covariance matrix. Modification is done with an algorithm based on linear algebra that minimizes changes to the means and covariances while insuring that the operations that impose the different constraints do not conflict with each other.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗