Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Datasets”

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 1,567 records · Page 87

SU-E-J-173: Evaluation of Deformable Registration for Correcting Respiratory Motion in 4DCT Lung Images

Purpose: Deformable Image Registration (DIR) is gaining wider clinical acceptance in radiation oncology. The aim of this work is to characterise a DIR algorithm on publically available 4DCT lung images, such that comparison can be performed against other algorithms. We propose an evaluation method of registration accuracy that takes into account the initial misregistration of the datasets. Methods: The “DIR Validation dataset” ( http://www.creatis.insa-lyon.fr/rio/dir{sub v}alidation{sub d}ata ) provides benchmark data for evaluating 3D CT registration algorithms. It consists of six 4DCT lung datasets (1x1x2mm resolution) with 100 landmarks identified on the end-exhalation and end-inhalation phases. Images were registered to end-inhalation using proprietary form of optical flow in commercial software (Mirada RTx, Mirada Medical, UK). Target registration error was measured before and after DIR, referred to as Initial Registration Error (IRE) and Final Registration Error (FRE). Results: The mean FRE over all landmarks was 1.37±1.81mm. FRE increased with IRE. Mean FRE of 0.86, 0.86, 1.53, 3.38, 4.45, 7.58mm was observed for IRE in the ranges 0–5, 5–10, 10–15, 15–20, 20–25, >25 mm. Higher FRE was observed at the inferior lung, where IRE was greater. Out-of-plane motion contributed more to IRE, and therefore to FRE. Maximum FRE of 20.6mm was observed for IRE of 32.1mm, located at the posterior of the middle lobe for dataset 2. Sub-voxel registration accuracy was achieved for up to 10mm IRE, and increased linearly at 0.3mm FRE/mm IRE thereafter. Conclusion: Publicly available clinical datasets enable algorithms to be compared objectively between publications. However, only reporting average TRE after registration can be misleading as the ability of an algorithm to correct for displacements varies with the IRE or position within the patient. Consequently, algorithms should be characterized using the entire range of initial displacements. For the algorithm assessed, clinically acceptable error within one voxel was achieved for IRE of up to 15mm. TK, AL, and MG are employees of Mirada Medical.

60 APPLIED LIFE SCIENCES↗

TH-A-9A-01: Active Optical Flow Model: Predicting Voxel-Level Dose Prediction in Spine SBRT

Purpose: To predict voxel-level dose distribution and enable effective evaluation of cord dose sparing in spine SBRT. Methods: We present an active optical flow model (AOFM) to statistically describe cord dose variations and train a predictive model to represent correlations between AOFM and PTV contours. Thirty clinically accepted spine SBRT plans are evenly divided into training and testing datasets. The development of predictive model consists of 1) collecting a sequence of dose maps including PTV and OAR (spinal cord) as well as a set of associated PTV contours adjacent to OAR from the training dataset, 2) classifying data into five groups based on PTV's locations relative to OAR, two “Top”s, “Left”, “Right”, and “Bottom”, 3) randomly selecting a dose map as the reference in each group and applying rigid registration and optical flow deformation to match all other maps to the reference, 4) building AOFM by importing optical flow vectors and dose values into the principal component analysis (PCA), 5) applying another PCA to features of PTV and OAR contours to generate an active shape model (ASM), and 6) computing a linear regression model of correlations between AOFM and ASM.When predicting dose distribution of a new case in the testing dataset, the PTV is first assigned to a group based on its contour characteristics. Contour features are then transformed into ASM's principal coordinates of the selected group. Finally, voxel-level dose distribution is determined by mapping from the ASM space to the AOFM space using the predictive model. Results: The DVHs predicted by the AOFM-based model and those in clinical plans are comparable in training and testing datasets. At 2% volume the dose difference between predicted and clinical plans is 4.2±4.4% and 3.3±3.5% in the training and testing datasets, respectively. Conclusion: The AOFM is effective in predicting voxel-level dose distribution for spine SBRT. Partially supported by NIH/NCI under grant #R21CA161389 and a master research grant by Varian Medical System.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

TH-C-BRD-05: Reducing Proton Beam Range Uncertainty with Patient-Specific CT HU to RSP Calibrations Based On Single-Detector Proton Radiography

Purpose: Beam range uncertainty in proton treatment comes primarily from converting the patient's X-ray CT (xCT) dataset to relative stopping power (RSP). Current practices use a single curve for this conversion, produced by a stoichiometric calibration based on tissue composition data for average, healthy, adult humans, but not for the individual in question. Proton radiographs produce water-equivalent path length (WEPL) maps, dependent on the RSP of tissues within the specific patient. This work investigates the use of such WEPL maps to optimize patient-specific calibration curves for reducing beam range uncertainty. Methods: The optimization procedure works on the principle of minimizing the difference between the known WEPL map, obtained from a proton radiograph, and a digitally-reconstructed WEPL map (DRWM) through an RSP dataset, by altering the calibration curve that is used to convert the xCT into an RSP dataset. DRWMs were produced with Plastimatch, an in-house developed software, and an optimization procedure was implemented in Matlab. Tests were made on a range of systems including simulated datasets with computed WEPL maps and phantoms (anthropomorphic and real biological tissue) with WEPL maps measured by single detector proton radiography. Results: For the simulated datasets, the optimizer showed excellent results. It was able to either completely eradicate or significantly reduce the root-mean-square-error (RMSE) in the WEPL for the homogeneous phantoms (to zero for individual materials or from 1.5% to 0.2% for the simultaneous optimization of multiple materials). For the heterogeneous phantom the RMSE was reduced from 1.9% to 0.3%. Conclusion: An optimization procedure has been designed to produce patient-specific calibration curves. Test results on a range of systems with different complexities and sizes have been promising for accurate beam range control in patients. This project was funded equally by the Engineering and Physical Sciences Research Council (UK) and Ion Beam Applications (Louvain-La-Neuve, Belgium)

07 ISOTOPE AND RADIATION SOURCES↗

SU-C-BRA-04: Use of Esophageal Wall Thickness in Evaluation of the Response to Chemoradiation Therapy for Esophageal Cancer

Purpose: To quantitatively evaluate the esophageal cancer response to chemoradiation therapy (CRT) by measuring the esophageal wall thickness in CT. Method: Two datasets were used in this study. The first dataset is composed of CT scans of 15 esophageal cancer patients and 15 normal controls. The second dataset is composed of 20 esophageal cancer patients who underwent PET/CT scans before (Pre-CRT) and after CRT (Post-CRT). We first segmented the esophagus using a multi-atlas-based algorithm. The esophageal wall thickness was then computed, on each slice, as the equivalent circle radius of the segmented esophagus excluding the lumen. To evaluate the changes of wall thickness, we computed the standard deviation (SD), coefficient of variation (COV, SD/Mean), and flatness [(Max–Min)/Mean] of wall thickness along the entire esophagus. Results: For the first dataset, the mean wall thickness of cancer patients and normal controls were 6.35 mm and 6.03 mm, respectively. The mean SD, COV, and flatness of the wall thickness were 2.59, 0.21, and 1.27 for the cancer patients and 1.99, 0.16, and 1.13 for normal controls. Statistically significant differences (p < 0.05) were identified in SD and flatness. For the second dataset, the mean wall thickness of pre-CRT and post-CRT patients was 7.13 mm and 6.84 mm, respectively. The mean SD, COV, and flatness were 1.81, 0.26, and 1.06 for pre-CRT and 1.69, 0.26, and 1.06 for post-CRT. Statistically significant difference was not identified for these measurements. Current results are based on the entire esophagus. We believe significant differences between pre- and post-CRT scans could be obtained, if we conduct the measurements at tumor sites. Conclusion: Results show thicker wall thickness in pre-CRT scans and differences in wall thickness changes between normal and abnormal esophagus. This demonstrated the potential of esophageal wall thickness as a marker in the tumor CRT response evaluation. This work was supported in part by the National Cancer Institute Grant R01CA172638.

60 APPLIED LIFE SCIENCES↗

SU-E-T-333: Dosimetric Impact of Rotational Error On the Target Coverage in IMPT Lung Cancer Plans

Purpose: The main purpose of this study was to investigate the impact of rotational (yaw, roll, and pitch) error on the planning target volume (PTV) coverage in lung cancer plans generated by intensity modulated proton therapy (IMPT). Methods: In this retrospective study, computed tomography (CT) dataset of previously treated lung case was used. IMPT plan were generated on the original CT dataset using left-lateral (LL) and posterior-anterior (PA) beams for a total dose of 74 Gy[RBE] with 2 Gy[RBE] per fraction. In order to investigate the dosimetric impact of rotational error, 12 new CT datasets were generated by re-sampling the original CT dataset for rotational (roll, yaw, and pitch) angles ranged from −5° to +5°, with an increment of 2.5°. A total of 12 new IMPT plans were generated based on the re-sampled CT datasets using beam parameters identical to the ones in the original IMPT plan. All treatment plans were generated in XiO treatment planning system. The PTV coverage (i.e., dose received by 95% of the PTV volume, D95) in new IMPT plans were then compared with the PTV coverage in the original IMPT plan. Results: Rotational errors caused the reduction in the PTV coverage in all 12 new IMPT plans when compared to the original IMPT lung plan. Specifically, the PTV coverage was reduced by 4.94% to 50.51% for yaw, by 4.04% to 23.74% for roll, and by 5.21% to 46.88% for pitch errors. Conclusion: Unacceptable dosimetric results were observed in new IMPT plans as the PTV coverage was reduced by up to 26.87% and 50.51% for rotational error of 2.5° and 5°, respectively. Further investigation is underway in evaluating the PTV coverage loss in the IMPT lung cancer plans for smaller rotational angle change.

60 APPLIED LIFE SCIENCES↗

WE-AB-204-08: Uncertainties in Kinetic Modeling of Shortened Dynamic 18F-FMISO PET Due to Image Noise

Purpose: To assess uncertainties in voxel-wise kinetic modeling (KM) of dynamic 18F-FMISO PET (dPET) due to image noise for different lengths of dPET acquisitions. Methods: 12 tumor time activity curves (TACs) deduced from 6 head and neck cancer patient dPET datasets (45min dPET + 10min frames at ∼95min and ∼160min respectively) were modeled using an irreversible two-tissue compartment model to estimate kinetic rate constants (KRCs) k3, vB, K1, and K1/k2 (reference standard). For each modeled TAC, 1000 noisy TACs were simulated by adding Gaussian noise with standard deviation equivalent to that observed on a voxel level in GE DSTE PET-CT. KM was conducted for each set of noisy TACs using (i) full dataset (170min), and repeated using (ii) 105min and (iii) 45min shortened subsets, with an input function that was image-derived from the dPET dataset of matching length. Absolute value of percent difference between KRCs from noisy TACs as estimated from either of 3 datasets and reference standard KRCs was used to represent bias. Results: For all KRCs, bias was higher for shortened datasets. For k3, bias was lower when actual k3 was higher: from 7%±5% (170min), 9%±7% (105min), and 14%±11% (45min) (actual k3=0.0098), to 32%±25% (170min), 45%±34% (95min), and 79%±48% (45min) (actual k3=0.0008). Similar relationship was observed for vB, with bias being between ∼7% (actual vB=0.28) and ∼26–34% (actual vB=0.08). For K1 [K1/k2], bias was ∼6–10% [∼2–3%] (dependence on actual value of K1 [K1/k2] was not observed). Conclusion: Uncertainties in voxel-wise KM of shortened 18F-FMISO dPET due to image noisy only are larger for shorter acquisitions and, for k3 [vB], the bias was found to be inversely correlated with the actual values of k3 [vB]. Using 45min subset, k3, vB, K1, and K1/k2 could be estimated to within ∼15–80%, ∼10–35%, ∼10%, and ∼3%, respectively.

60 APPLIED LIFE SCIENCES↗

WE-G-207-06: 3D Fluoroscopic Image Generation From Patient-Specific 4DCBCT-Based Motion Models Derived From Physical Phantom and Clinical Patient Images

Purpose: Respiratory-correlated cone-beam CT (4DCBCT) images acquired immediately prior to treatment have the potential to represent patient motion patterns and anatomy during treatment, including both intra- and inter-fractional changes. We develop a method to generate patient-specific motion models based on 4DCBCT images acquired with existing clinical equipment and used to generate time varying volumetric images (3D fluoroscopic images) representing motion during treatment delivery. Methods: Motion models are derived by deformably registering each 4DCBCT phase to a reference phase, and performing principal component analysis (PCA) on the resulting displacement vector fields. 3D fluoroscopic images are estimated by optimizing the resulting PCA coefficients iteratively through comparison of the cone-beam projections simulating kV treatment imaging and digitally reconstructed radiographs generated from the motion model. Patient and physical phantom datasets are used to evaluate the method in terms of tumor localization error compared to manually defined ground truth positions. Results: 4DCBCT-based motion models were derived and used to generate 3D fluoroscopic images at treatment time. For the patient datasets, the average tumor localization error and the 95th percentile were 1.57 and 3.13 respectively in subsets of four patient datasets. For the physical phantom datasets, the average tumor localization error and the 95th percentile were 1.14 and 2.78 respectively in two datasets. 4DCBCT motion models are shown to perform well in the context of generating 3D fluoroscopic images due to their ability to reproduce anatomical changes at treatment time. Conclusion: This study showed the feasibility of deriving 4DCBCT-based motion models and using them to generate 3D fluoroscopic images at treatment time in real clinical settings. 4DCBCT-based motion models were found to account for the 3D non-rigid motion of the patient anatomy during treatment and have the potential to localize tumor and other patient anatomical structures at treatment time even when inter-fractional changes occur. This project was supported, in part, through a Master Research Agreement with Varian Medical Systems, Inc., Palo Alto, CA. The project was also supported, in part, by Award Number R21CA156068 from the National Cancer Institute.

60 APPLIED LIFE SCIENCES↗

Quantifying the accuracy of the tumor motion and area as a function of acceleration factor for the simulation of the dynamic keyhole magnetic resonance imaging method

Purpose: The dynamic keyhole is a new MR image reconstruction method for thoracic and abdominal MR imaging. To date, this method has not been investigated with cancer patient magnetic resonance imaging (MRI) data. The goal of this study was to assess the dynamic keyhole method for the task of lung tumor localization using cine-MR images reconstructed in the presence of respiratory motion. Methods: The dynamic keyhole method utilizes a previously acquired a library of peripheral k-space datasets at similar displacement and phase (where phase is simply used to determine whether the breathing is inhale to exhale or exhale to inhale) respiratory bins in conjunction with central k-space datasets (keyhole) acquired. External respiratory signals drive the process of sorting, matching, and combining the two k-space streams for each respiratory bin, thereby achieving faster image acquisition without substantial motion artifacts. This study was the first that investigates the impact of k-space undersampling on lung tumor motion and area assessment across clinically available techniques (zero-filling and conventional keyhole). In this study, the dynamic keyhole, conventional keyhole and zero-filling methods were compared to full k-space dataset acquisition by quantifying (1) the keyhole size required for central k-space datasets for constant image quality across sixty four cine-MRI datasets from nine lung cancer patients, (2) the intensity difference between the original and reconstructed images in a constant keyhole size, and (3) the accuracy of tumor motion and area directly measured by tumor autocontouring. Results: For constant image quality, the dynamic keyhole method, conventional keyhole, and zero-filling methods required 22%, 34%, and 49% of the keyhole size (P < 0.0001), respectively, compared to the full k-space image acquisition method. Compared to the conventional keyhole and zero-filling reconstructed images with the keyhole size utilized in the dynamic keyhole method, an average intensity difference of the dynamic keyhole reconstructed images (P < 0.0001) was minimal, and resulted in the accuracy of tumor motion within 99.6% (P < 0.0001) and the accuracy of tumor area within 98.0% (P < 0.0001) for lung tumor monitoring applications. Conclusions: This study demonstrates that the dynamic keyhole method is a promising technique for clinical applications such as image-guided radiation therapy requiring the MR monitoring of thoracic tumors. Based on the results from this study, the dynamic keyhole method could increase the imaging frequency by up to a factor of five compared with full k-space methods for real-time lung tumor MRI.

60 APPLIED LIFE SCIENCES↗

Functional analysis of metagenomes and metatranscriptomes using SEED and KEGG

Background: Metagenomics is the study of microbial organisms using sequencing applied directly to environmental samples. Technological advances in next-generation sequencing methods are fueling a rapid increase in the number and scope of metagenome projects. While metagenomics provides information on the gene content, metatranscriptomics aims at understanding gene expression patterns in microbial communities. The initial computational analysis of a metagenome or metatranscriptome addresses three questions: (1) Who is out there? (2) What are they doing? and (3) How do different datasets compare? There is a need for new computational tools to answer these questions. In 2007, the program MEGAN (MEtaGenome ANalyzer) was released, as a standalone interactive tool for analyzing the taxonomic content of a single metagenome dataset. The program has subsequently been extended to support comparative analyses of multiple datasets. Results: The focus of this paper is to report on new features of MEGAN that allow the functional analysis of multiple metagenomes (and metatranscriptomes) based on the SEED hierarchy and KEGG pathways. We have compared our results with the MG-RAST service for different datasets. Conclusions: The MEGAN program now allows the interactive analysis and comparison of the taxonomical and functional content of multiple datasets. As a stand-alone tool, MEGAN provides an alternative to web portals for scientists that have concerns about uploading their unpublished data to a website.

59 BASIC BIOLOGICAL SCIENCES↗

VA Community Determinants of Health Data Curation Documentation FY26-Q2

The U.S. Department of Veterans Affairs (VA) places the health and well-being of our nation’s veterans as its top priority. VA is dedicated to offering timely access to high-quality, evidence-based mental health care that meets the needs of veterans and supports their reintegration into society. One of our core missions is to prevent suicide among veterans through innovative approaches and resources. With funding from the VA Office of Mental Health and Suicide Prevention (OMHSP), the Community Determinants of Health (EDH) Data project has developed innovative datasets associated with specific health outcomes, a methodology for transforming spatiotemporal data from one spatial reference (e.g., a 1 km grid) to another (e.g., U.S. Census Tracts), and capabilities for modeling health outcomes. These datasets represent an enhancement of the Agency for Healthcare Research and Quality (AHRQ), addressing key gaps by introducing finer spatial resolution (Census Tract) and additional geographical covariates into existing data. The curation and standardization of these datasets is a complex task since they often originate from various sources and are measured at different spatial and temporal resolutions. For example, U.S. Census data products typically use census blocks, block groups, or counties, while data like weather data are available on 1 km grids. Some economic data may only be available at the ZIP code level. In this context, ‘standardized’ means that all datasets share the same spatial extent (e.g., U.S. Census Tract and/or County), and ‘curated’ implies a repeatable process with data provenance and the use of appropriate methodologies for covariate conversion. The Community Determinants of Health datasets draw from multiple sources, resulting in variables with varying degrees of availability, patterns of missing data, and methodological considerations across different sources, geographies, and years.

99 GENERAL AND MISCELLANEOUS↗

Worldwide Population Estimates for Small Geographic Areas

This chapter discusses the basis for these population estimates, scope, and limitations based on experiences in development of massive global population datasets and usage of these datasets as a basis for sampling design. It presents tools and approaches for using these georeferenced population estimates for complex household survey sampling. The chapter provides a unified resource for understanding current gridded population datasets, their use in survey research, and promising areas of future work toward improved population estimates. It includes an overview of popular gridded population datasets, common methodological frameworks for developing gridded population estimates, and discusses pros and cons for selecting a gridded population dataset from a survey perspective. The chapter focuses on survey methods that use gridded population estimates specifically. It highlights differences between census population estimates and gridded population estimates that may be relevant when planning data collection. The chapter presents a case study of gridded population data and sampling methods in Nigeria.

Amer, Safaa↗

The Twentieth Century Reanalysis Project

The Twentieth Century Reanalysis (20CR) project is an international effort to produce a comprehensive global atmospheric circulation dataset spanning the twentieth century, assimilating only surface pressure reports and using observed monthly sea-surface temperature and sea-ice distributions as boundary conditions. It is chiefly motivated by a need to provide an observational dataset with quantified uncertainties for validations of climate model simulations of the twentieth century onall time-scales,with emphasisonthe statistics of daily weather. It uses an Ensemble Kalman Filter data assimilation method with background ‘first guess’ fields supplied by an ensemble of forecasts from a global numerical weather prediction model. This directly yields a global analysis every 6 hours as the most likely state of the atmosphere, and also an uncertainty estimate of that analysis. The 20CR dataset provides the first estimates of global tropospheric variability, and of the dataset’s time-varying quality, from 1871 to the present at 6-hourly temporal and 2° spatial resolutions. Intercomparisons with independent radiosonde data indicate that the reanalyses are generally of high quality. The quality in the extratropical Northern Hemisphere throughout the century is similar to that of current three-day operational NWP forecasts. Intercomparisons over the second half-century of these surface-based reanalyses with other reanalyses that also make use of upper-air and satellite data are equally encouraging. It is anticipated that the 20CR dataset will be a valuable resource to the climate research community for both model validations and diagnostic studies. Some surprising results are already evident. For instance, the long-term trends of indices representing the North Atlantic Oscillation, the tropical Pacific Walker Circulation, and the Pacific–North American pattern are weak or non-existent over the full period of record. The long-term trends of zonally averaged precipitation minus evaporation also differ in character from those in climate model simulations of the twentieth century.

Meteorology & Atmospheric Sciences↗

Increased inflammation as well as decreased endoplasmic reticulum stress and translation differentiate pancreatic islets from donors with pre-symptomatic stage 1 type 1 diabetes and non-diabetic donors

Aims/hypothesis Progression to type 1 diabetes is associated with genetic factors, the presence of autoantibodies and a decline in beta cell insulin secretion in response to glucose. Very little is known regarding the molecular changes that occur in human insulin-secreting beta cells prior to the onset of type 1 diabetes. Herein, we applied an unbiased proteomics approach to identify changes in proteins and potential mechanisms of islet dysfunction in islet-autoantibody-positive organ donors with pre-symptomatic stage 1 type 1 diabetes (HbA1c ≤42 mmol/mol [6.0%]). We aimed to identify pathways in islets that are indicative of beta cell dysfunction. Methods Multiple islet sections were collected through laser microdissection of frozen pancreatic tissues from organ donors positive for single or multiple islet autoantibodies (AAb + , n=5), and age (±2 years)- and sex-matched non-diabetic (ND) control donors (n=5) obtained from the Network for Pancreatic Organ donors with Diabetes (nPOD). Islet sections were subjected to MS-based proteomics and analysed with label-free quantification followed by pathway and functional annotations. Results Analyses resulted in ~4500 proteins identified with low false discovery rate (<1%), with 2165 proteins reliably quantified in every islet sample. We observed large inter-donor variations that presented a challenge for statistical analysis of proteome changes between donor groups. We therefore focused on only the donors with stage 1 type 1 diabetes who were positive for multiple autoantibodies (mAAb + , n=3) and genetic risk compared with their matched ND controls (n=3) for the final statistical analysis. Approximately 10% of the proteins (n=202) were significantly different (unadjusted p<0.025, q<0.15) for mAAb + vs ND donor islets. The significant alterations clustered around major functions for upregulation in the immune response and glycolysis, and downregulation in endoplasmic reticulum (ER) stress response as well as protein translation and synthesis. The observed proteome changes were further supported by several independent published datasets, including a proteomics dataset from in vitro proinflammatory cytokine-treated human islets and single-cell RNA-seq datasets from AAb + individuals. Conclusions/interpretation In situ human islet proteome alterations in stage 1 type 1 diabetes centred around several major functional categories, including an expected increase in immune response genes (elevated antigen presentation/HLA), with decreases in protein synthesis and ER stress response, as well as compensatory metabolic response. The dataset serves as a proteomics resource for future studies on beta cell changes during type 1 diabetes progression and pathogenesis. Data availability The LC-MS raw datasets that support the findings of this study have been deposited in the online repository: MassIVE (https://massive.ucsd.edu/ProteoSAFe/static/massive.jsp) with accession no. MSV000090212.

Autoantibody-positive↗

Estimation of split renal function using different volumetric methods: inter- and intraindividual comparison between MRI and CT

Purpose: This study aims to determine whether contrast-enhanced (CE)-magnetic resonance imaging (MRI) is comparable to CE-computed tomography (CT) for estimation of split renal function (SRF). For this purpose, two different kidney volumetry methods, the renal cortex volumetry (RCV) and modified ellipsoid volume (MELV), are compared for both acquisition types (CT vs. MRI) with regard to accuracy and reliability, subsequently referred to as RCV{sub CT}/RCV{sub MRI} and MELV{sub CT}/MELV{sub MRI}. Methods: This retrospective study included 29 patients (18 men and 11 women; mean age 62.8 ± 12.4 years) who underwent CE-MRI and CE-CT of the abdomen within a period of 3 months. Two independent readers (R1/R2) performed RCV and MELV in all datasets with corresponding semiautomated software tools. RCV was performed with datasets in the arterial phase and MELV in the venous phase. Statistics were calculated using one-way ANOVA, two-tailed Student’s t test, Pearson´s correlation, and Bland–Altman plots with p ≤ 0.05 being considered statistically significant. Results: In all datasets, SRF was almost identical for both volumetry methods with a mean difference of < 1%. Bland–Altman analysis comparing RCV in CT and MRI showed very good agreement for R1/R2. Interreader agreement was strong for RCV{sub CT} and good for RCV{sub MRI} (r = 0.89; r = 0.69). MELV{sub CT/MRI} interreader agreement was only moderate (r = 0.54; r = 0.50) with a high range of values. Intrareader agreement was excellent for all measurements, except MELV{sub MRI} which showed a high mean bias and range of values (RCV{sub CT}: r = 0.93, RCV{sub MRI}: r = 0.98, MELV{sub CT}: r = 0.89, MELV{sub MRI}: r = 0.54). Conclusion: Renal volumetric estimates of SRF are almost as accurate and reliable with CE-MRI as with CE-CT using RCV method. In distinction, the calculation of SRF using MELV was inferior to RCV with respect to accuracy and reliability. Thus, RCV method is recommended to estimate SRF, primarily using CT datasets. However, RCV with MRI datasets for kidney volumetry allows for comparable accuracy and reliability while sparing patients and healthy donors of unnecessary radiation exposure.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Total Blood Mercury Predicts Methylmercury Exposure in Fish and Shellfish Consumers

Abstract Many studies evaluating methylmercury (MeHg) toxicity rely on whole blood total mercury (THg) measurements to estimate MeHg exposure. However, whole blood THg includes other forms of mercury (Hg), such as inorganic Hg, which have different exposure sources and toxicological effects than MeHg. Therefore, estimating the whole blood MeHg/THg ratio is critical to predicting MeHg exposure and, subsequently, efforts to establish an exposure–response relationship for use in risk assessment. A large, representative dataset (National Health and Nutrition Examination Survey (NHANES) 2011–2016) was used to determine the whole blood MeHg/THg ratio among (a) self-reported fish and shellfish consumers, ≥ 15 years of age (the “full adult” population ( N = 5268 training dataset; N = 2336 test dataset)) and (b) female fish and shellfish consumers, 15–44 years of age (the “women of reproductive age” population ( N = 1285 training dataset; N = 560 test dataset)). Unadjusted and adjusted linear and spline models with direct measurements for both THg and MeHg were evaluated. The mean whole blood MeHg/THg ratio was 0.75 (95% confidence interval (CI): 0.74, 0.75). This ratio was significantly higher among those with higher THg concentrations. All models exhibited excellent fit (adjusted R 2 from 0.957 to 0.982). Performance was slightly improved in spline versus linear models. For the full adult population and women of reproductive age, the unadjusted spline model predicted whole blood MeHg concentrations of 5.65 µg/L and 5.55 µg/L, respectively, when the THg concentration was 5.80 µg/L. These results suggest that whole blood THg is a good predictor of whole blood MeHg among fish and shellfish consumers.

Wells, Ellen M.↗

A structural equation modeling approach to leveraging the power of extant sentiment analysis tools

Machine-derived sentiment analysis has become a pervasive and useful tool to address a wide array of issues in natural language processing. Leading technology companies such as Google now provide sentiment analysis tools (SATs) as readily accessible online products. Academic researchers develop and make available SATs to support the research enterprise. One of the major challenges with SATs is the inconsistencies in results among the various SATs. Consequently, the selection of a SAT for a specific purpose may significantly impact the application. This study addresses the foregoing problem by utilizing structural equation modeling to merge the outputs of SATs to develop a combined sentiment metric without the need for a labeled training dataset. This method is applicable to a wide range of text-based problems, is data-driven, and replicable. It was tested using three publicly available datasets and compared against seven different SATs. The results indicate that as a continous measure, the proposed method outperformed other SATs in the movie reviews and SemEval datasets, and achieved a tie for first place with IBM Watson on the Sentiment 140 dataset. Also, compared to the published major alternatives, the arithmetic mean solution, this approach performed better across these three datasets.

97 MATHEMATICS AND COMPUTING↗

Large-scale seismic waveform quality metric calculation using Hadoop

Here in this work we investigated the suitability of Hadoop MapReduce and Apache Spark for large-scale computation of seismic waveform quality metrics by comparing their performance with that of a traditional distributed implementation. The Incorporated Research Institutions for Seismology (IRIS) Data Management Center (DMC) provided 43 terabytes of broadband waveform data of which 5.1 TB of data were processed with the traditional architecture, and the full 43 TB were processed using MapReduce and Spark. Maximum performance of ~0.56 terabytes per hour was achieved using all 5 nodes of the traditional implementation. We noted that I/O dominated processing, and that I/O performance was deteriorating with the addition of the 5th node. Data collected from this experiment provided the baseline against which the Hadoop results were compared. Next, we processed the full 43 TB dataset using both MapReduce and Apache Spark on our 18-node Hadoop cluster. We conducted these experiments multiple times with various subsets of the data so that we could build models to predict performance as a function of dataset size. We found that both MapReduce and Spark significantly outperformed the traditional reference implementation. At a dataset size of 5.1 terabytes, both Spark and MapReduce were about 15 times faster than the reference implementation. Furthermore, our performance models predict that for a dataset of 350 terabytes, Spark running on a 100-node cluster would be about 265 times faster than the reference implementation. We do not expect that the reference implementation deployed on a 100-node cluster would perform significantly better than on the 5-node cluster because the I/O performance cannot be made to scale. Finally, we note that although Big Data technologies clearly provide a way to process seismic waveform datasets in a high-performance and scalable manner, the technology is still rapidly changing, requires a high degree of investment in personnel, and will likely require significant changes in other parts of our infrastructure. Nevertheless, we anticipate that as the technology matures and third-party tool vendors make it easier to manage and operate clusters, Hadoop (or a successor) will play a large role in our seismic data processing.

58 GEOSCIENCES↗

Machine learning assisted phase and size-controlled synthesis of iron oxide particles

Synthesis of iron oxides with specific phases and particle sizes is a crucial challenge in various fields, including materials science, energy storage, biomedical applications, environmental science, and earth science. However, despite significant advances in this area, much of the current palette of particle outcomes has been based on time-consuming trial-and-error exploration of synthesis conditions. The present study was designed to explore a very different approach to 1) predict the outcome of synthesis from specified reaction parameters based on using machine learning (ML) techniques, and 2) correlate sets of parameters to obtain products with desired outcomes by a newly designed recommendation algorithm. To achieve this, four ML algorithms were tested, namely random forest, logistic regression, support vector machine, and k-nearest neighbor. Among the models, random forest outperformed the others, attaining 96% and 81% accuracy when predicting the phase and size of iron oxide particles in the test dataset. Surprisingly, the permutation feature importance analysis revealed that volume, which may strongly relate to pressure, was one of the important features, along with precursor concentration, pH, temperature, and time, influencing the phase and size of iron oxide particles during synthesis. To verify the robustness of the random forest models, prediction and experimental results were compared based on 24 randomly generated methods in additive and non-additive systems not included in the datasets. The predictions of product phase and particle size from the models agreed well with the experimental results. Furthermore, a searching and ranking algorithm was developed to recommend potential synthesis parameters for obtaining iron oxide products with the desired phase and particle size from previous studies in the dataset. Furthermore, this study lays the foundation for a closed-loop approach in materials synthesis and preparation, beginning with suggesting potential reaction parameters from the dataset and predicting potential outcomes, followed by conducting experiments and analyses, and ultimately enriching the dataset.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗