Search NASA⌕ Search

SEARCH · Search NASA

Results for “Automatic Labeling”

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 145 records · Page 8

Exudate-based diabetic macular edema detection in fundus images using publicly available datasets

Diabetic macular edema (DME) is a common vision threatening complication of diabetic retinopathy. In a large scale screening environment DME can be assessed by detecting exudates (a type of bright lesions) in fundus images. In this work, we introduce a new methodology for diagnosis of DME using a novel set of features based on colour, wavelet decomposition and automatic lesion segmentation. These features are employed to train a classifier able to automatically diagnose DME through the presence of exudation. We present a new publicly available dataset with ground-truth data containing 169 patients from various ethnic groups and levels of DME. This and other two publicly available datasets are employed to evaluate our algorithm. We are able to achieve diagnosis performance comparable to retina experts on the MESSIDOR (an independently labelled dataset with 1200 images) with cross-dataset testing (e.g., the classifier was trained on an independent dataset and tested on MESSIDOR). Our algorithm obtained an AUC between 0.88 and 0.94 depending on the dataset/features used. Additionally, it does not need ground truth at lesion level to reject false positives and is computationally efficient, as it generates a diagnosis on an average of 4.4 s (9.3 s, considering the optic nerve localization) per image on an 2.6 GHz platform with an unoptimized Matlab implementation.

59 BASIC BIOLOGICAL SCIENCES↗

Situational Awareness of Network System Roles (SANSR)

In a large enterprise it is difficult for cyber security analysts to know what services and roles every machine on the network is performing (e.g. file server, domain name server, email server). Understanding the roles of the systems in the network provides analysts with a situational awareness that will allow them to detect consequential changes in the network, initiate an incident response plan, and optimize their security posture. Using the network flow data, already collected by most enterprises, we developed a tool that enables analysts to automatically detect/classify services and roles of every machine that’s operating on a network (e.g. file server, domain name server, email server) for better situational awareness of potential threats to the network. his tool queries Elasticsearch for network flow data, creates a temporal behavior model of each system, uses unsupervised machine learning to cluster the models with a set of labeled temporal behavior models, and the resulting information can be printed to the console or programmatically accessed. The results include the likelihood that a machine has a labeled role and lists other machines that are most similar in behavior.

Huffer, KellyM.↗

Constraining Genome-Scale Models to Represent the Bow Tie Structure of Metabolism for 13 C Metabolic Flux Analysis

Determination of internal metabolic fluxes is crucial for fundamental and applied biology because they map how carbon and electrons flow through metabolism to enable cell function. 13 C Metabolic Flux Analysis ( 13 C MFA) and Two-Scale 13 C Metabolic Flux Analysis (2S- 13 C MFA) are two techniques used to determine such fluxes. Both operate on the simplifying approximation that metabolic flux from peripheral metabolism into central “core” carbon metabolism is minimal, and can be omitted when modeling isotopic labeling in core metabolism. The validity of this “two-scale” or “bow tie” approximation is supported both by the ability to accurately model experimental isotopic labeling data, and by experimentally verified metabolic engineering predictions using these methods. However, the boundaries of core metabolism that satisfy this approximation can vary across species, and across cell culture conditions. Here, we present a set of algorithms that (1) systematically calculate flux bounds for any specified “core” of a genome-scale model so as to satisfy the bow tie approximation and (2) automatically identify an updated set of core reactions that can satisfy this approximation more efficiently. First, we leverage linear programming to simultaneously identify the lowest fluxes from peripheral metabolism into core metabolism compatible with the observed growth rate and extracellular metabolite exchange fluxes. Second, we use Simulated Annealing to identify an updated set of core reactions that allow for a minimum of fluxes into core metabolism to satisfy these experimental constraints. Together, these methods accelerate and automate the identification of a biologically reasonable set of core reactions for use with 13 C MFA or 2S- 13 C MFA, as well as provide for a substantially lower set of flux bounds for fluxes into the core as compared with previous methods. We provide an open source Python implementation of these algorithms at https://github.com/JBEI/limitfluxtocore.

13C metabolic flux analysis↗

Large-Scale Solar Panel Mapping from Aerial Images Using Deep Convolutional Networks

Up-to-date maps of installed solar photovoltaic panels are a critical input for policy and financial assessment of solar distributed generation. However, such maps for large areas are not available. With high coverage and low cost, aerial images enable large-scale mapping, bit it is highly difficult to automatically identify solar panels from images, which are small objects with varying appearances dispersed in complex scenes. We introduce a new approach based on deep convolutional networks, which effectively learns to delineate solar panels in aerial scenes. The approach has successfully mapped solar panels in imagery covering 200 square kilometers in two cities, using only 12 square kilometers of training data that are manually labeled.

Yuan, Jiangye↗

PRMI: A Dataset of Minirhizotron Images for Diverse Plant Root Study

Understanding a plant's root system architecture (RSA) is crucial for a variety of plant science problem domains including sustainability and climate adaptation. Minirhizotron (MR) technology is a widely-used approach for phenotyping RSA non-destructively by capturing root imagery over time. Precisely segmenting roots from the soil in MR imagery is a critical step in studying RSA features. In this paper, we introduce a large-scale dataset of plant root images captured by MR technology. In total, there are over 72K RGB root images across six different species including cotton, papaya, peanut, sesame, sunflower, and switchgrass in the dataset. The images span a variety of conditions including varied root age, root structures, soil types, and depths under the soil surface. All of the images have been annotated with weak image-level labels indicating whether each image contains roots or not. The image-level labels can be used to support weakly supervised learning in plant root segmentation tasks. In addition, 63K images have been manually annotated to generate pixel-level binary masks indicating whether each pixel corresponds to root or not. These pixel-level binary masks can be used as ground truth for supervised learning in semantic segmentation tasks. By introducing this dataset, we aim to facilitate the automatic segmentation of roots and the research of RSA with deep learning and other image analysis algorithms.

Xu, Weihuang↗

Machine Learning Benchmarks for the Classification of Equivalent Circuit Models from Electrochemical Impedance Spectra

Analysis of Electrochemical Impedance Spectroscopy (EIS) data for electrochemical systems often consists of defining an Equivalent Circuit Model (ECM) using expert knowledge and then optimizing the model parameters to deconvolute various resistance, capacitive, inductive, or diffusion responses. For small data sets, this procedure can be conducted manually; however, it is not feasible to manually define a proper ECM for extensive data sets with a wide range of EIS responses. Automatic identification of an ECM would substantially accelerate the analysis of large sets of EIS data. We showcase machine learning methods to classify the ECMs of 9,300 impedance spectra provided by QuantumScape for the BatteryDEV hackathon. The best-performing approach is a gradient-boosted tree model utilizing a library to automatically generate features, followed by a random forest model using the raw spectral data. A convolutional neural network using boolean images of Nyquist representations is presented as an alternative, although it achieves a lower accuracy. We publish the data and open source the associated code. The approaches described in this article can serve as benchmarks for further studies. A key remaining challenge is the identifiability of the labels, underlined by the model performances and the comparison of misclassified spectra.

25 ENERGY STORAGE↗

Massive all-atom analysis of 2D materials with quantum properties (Final report)

Improvements in microscopy have enabled the acquisition of data at a scale that is difficult to process manually, making automated machine learning approaches to analyzing experimental images essential. In this project, we developed and applied machine learning (ML) workflows for atomic resolution scanning transmission electron microscopy (STEM) images. This development included improving both methodology as well as generating user-friendly codes. We developed machine learning architectures which, after training, automatically identify the location and types of defects throughout a material. We used these data to produce class-averaged images of 2D atomic coordinates with up to 0.3 pm precision, uncovering the structure and oscillations of long-range strain fields around point defects in WSe 2-2x Te 2x . We also resolved a long-standing problem in this field in the training of ML models, a lack of labeled experimental data, by developing a cycle-GAN that transformed simulated-generated labeled data into labeled data indistinguishable from experiment and therefore suitable for training. This removed the remaining parts of the ML data processing workflow where human intervention was still critical and therefore a bottleneck to working at scale. Codes have been developed and released for this full machine learning workflow. ML approaches to partially automate STEM acquisition were also developed. Finally we applied ML and other advanced data processing methods to several materials science problems in two-dimensional materials, including studying the evolution of hyperuniformity with defect concentration in WSe2, understanding phase transformations in transition metal dichalcogenides during in-situ heating in the STEM, and exploring how 2D interfaces transform from twisted into aligned structures.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

SU-F-R-25: Automatic Identification of Suspicious Recurrent/residual Disease Regions After Prostatectomy

Purpose: To automatically identify and outline suspicious regions of recurrent or residual disease in the prostate bed using Dynamic Contrast Enhanced-MRI (DCE-MRI) in patients after prostatectomy. Methods: Twenty-two patients presenting for salvage radiotherapy and with identified Gross Tumor Volume (GTV) in the prostate bed were retrospectively analyzed. The MRI data consisted of Axial T2weighted-MRI (T2w) of the pelvis: resolution 1.25×1.25×2.5 mm; Field of View (FOV): 320×320 mm; slice thickness=2.5mm; 72 slices; and Dynamic Contrast Enhanced MRI (DCE-MRI)–12 series of T1w with identical spatial resolution to T2w and at 30–34s temporal resolution. Unsupervised pattern recognition was used to decompose the 4D DCE data as the product W.H of weights W of k patterns H. A well-perfused pattern Hwp was identified and the weight map Wwp associated to Hwp was used to delineate suspicious volumes. Threshold of Wwp set at mean(Wwp)+S*std(Wwp), S=1,1.5,2 and 2.5 defined four volumes labeled as DCE1.0 to DCE2.5. These volumes were displayed on T2w and, along with GTV, were correlated with the highest pre-treatment PSA values, and with pharmacokinetic analysis constants. Results: GTV was significantly correlated with DCE2.0(ρ= 0.60, p<0.003), and DCE 2.5 (ρ=0.58, p=0.004)). Significant correlation was found between highest pre-treatment PSA and GTV(ρ=0.42, p<0.049), DCE2.0(ρ= 0.52, p<0.012), and DCE 2.5 (ρ=0.67, p<<0.01)). Kruskal-Wallis analysis showed that Ktrans median value was statistically different between non-specific prostate bed tissue NSPBT and both GTV (p<<0.001) and DCE2.5 (p<<0.001), but while median Ve was statistically different between DCE2.5 and NSPBT (p=0.002), it was not statistically different between GTV and NSPBT (p=0.054), suggesting that automatic volumes capture more accurately the area of malignancy. Conclusion: Software developed for identification and visualization of suspicions regions in DCE-MRI from post-prostatectomy patients has been validated by PSA and pharmacokinetic constants analysis showing that it generates clinically relevant volumes.

60 APPLIED LIFE SCIENCES↗

Automated bone segmentation from dental CBCT images using patch-based sparse representation and convex optimization

Purpose: Cone-beam computed tomography (CBCT) is an increasingly utilized imaging modality for the diagnosis and treatment planning of the patients with craniomaxillofacial (CMF) deformities. Accurate segmentation of CBCT image is an essential step to generate three-dimensional (3D) models for the diagnosis and treatment planning of the patients with CMF deformities. However, due to the poor image quality, including very low signal-to-noise ratio and the widespread image artifacts such as noise, beam hardening, and inhomogeneity, it is challenging to segment the CBCT images. In this paper, the authors present a new automatic segmentation method to address these problems. Methods: To segment CBCT images, the authors propose a new method for fully automated CBCT segmentation by using patch-based sparse representation to (1) segment bony structures from the soft tissues and (2) further separate the mandible from the maxilla. Specifically, a region-specific registration strategy is first proposed to warp all the atlases to the current testing subject and then a sparse-based label propagation strategy is employed to estimate a patient-specific atlas from all aligned atlases. Finally, the patient-specific atlas is integrated into amaximum a posteriori probability-based convex segmentation framework for accurate segmentation. Results: The proposed method has been evaluated on a dataset with 15 CBCT images. The effectiveness of the proposed region-specific registration strategy and patient-specific atlas has been validated by comparing with the traditional registration strategy and population-based atlas. The experimental results show that the proposed method achieves the best segmentation accuracy by comparison with other state-of-the-art segmentation methods. Conclusions: The authors have proposed a new CBCT segmentation method by using patch-based sparse representation and convex optimization, which can achieve considerably accurate segmentation results in CBCT segmentation based on 15 patients.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Defect detection in atomic-resolution images via unsupervised learning with translational invariance

Abstract Crystallographic defects can now be routinely imaged at atomic resolution with aberration-corrected scanning transmission electron microscopy (STEM) at high speed, with the potential for vast volumes of data to be acquired in relatively short times or through autonomous experiments that can continue over very long periods. Automatic detection and classification of defects in the STEM images are needed in order to handle the data in an efficient way. However, like many other tasks related to object detection and identification in artificial intelligence, it is challenging to detect and identify defects from STEM images. Furthermore, it is difficult to deal with crystal structures that have many atoms and low symmetries. Previous methods used for defect detection and classification were based on supervised learning, which requires human-labeled data. In this work, we develop an approach for defect detection with unsupervised machine learning based on a one-class support vector machine (OCSVM). We introduce two schemes of image segmentation and data preprocessing, both of which involve taking the Patterson function of each segment as inputs. We demonstrate that this method can be applied to various defects, such as point and line defects in 2D materials and twin boundaries in 3D nanocrystals.

36 MATERIALS SCIENCE↗

Automated Radiation Targeting in Head-and-Neck Cancer Using Region-Based Texture Analysis of PET and CT Images

Purpose: A co-registered multimodality pattern analysis segmentation system (COMPASS) was developed to automatically delineate the radiation targets in head-and-neck cancer (HNC) using both {sup 18}F-fluoro-deoxy glucose-positron emission tomography (PET) and computed tomography (CT) images. The performance of the COMPASS was compared with the results of existing threshold-based methods and radiation oncologist-drawn contours. Methods and Materials: The COMPASS extracted texture features from corresponding PET and CT voxels. Using these texture features, a decision-tree-based K-nearest-neighbor classifier labeled each voxel as either 'normal' or 'abnormal.' The COMPASS was applied to the PET/CT images of 10 HNC patients. Automated segmentation results were validated against the manual segmentations of three radiation oncologists using the volume, sensitivity, and specificity. The performance of the COMPASS was compared with three PET-based threshold methods: standard uptake value of 2.5, 50% maximal intensity, and signal/background ratio. Results: The tumor delineations of the COMPASS were both quantitatively and qualitatively more similar to those of the radiation oncologists than the delineations from the other methods. The specificity was 95% {+-} 2%, 84% {+-} 9%, 98% {+-} 3%, and 96% {+-} 4%, and the sensitivity was 90% {+-} 12%, 93% {+-} 10%, 48% {+-} 20%, and 68% {+-} 25% for the COMPASS, for a standard uptake value of 2.5, 50% maximal intensity, and signal/background ratio, respectively. The COMPASS distinguished HNC from adjacent normal tissues with high physiologic uptake and consistently defined tumors with large variability in {sup 18}F-fluoro-deoxy glucose uptake, which are often problematic with the threshold-based methods. Conclusion: Automated segmentation using texture analysis of PET/CT images has the potential to provide accurate delineation of HNC. This could lead to reduced interobserver variability, reduced uncertainty in target delineation, and improved treatment planning accuracy.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Characterizing the DNA damage response by cell tracking algorithms and cell features classification using high-content time-lapse analysis

Traditionally, the kinetics of DNA repair have been estimated using immunocytochemistry by labeling proteins involved in the DNA damage response (DDR) with fluorescent markers in a fixed cell assay. However, detailed knowledge of DDR dynamics across multiple cell generations cannot be obtained using a limited number of fixed cell time-points. Here we report on the dynamics of 53BP1 radiation induced foci (RIF) across multiple cell generations using live cell imaging of non-malignant human mammary epithelial cells (MCF10A) expressing histone H2B-GFP and the DNA repair protein 53BP1-mCherry. Using automatic extraction of RIF imaging features and linear programming techniques, we were able to characterize detailed RIF kinetics for 24 hours before and 24 hours after exposure to low and high doses of ionizing radiation. High-content-analysis at the single cell level over hundreds of cells allows us to quantify precisely the dose dependence of 53BP1 protein production, RIF nuclear localization and RIF movement after exposure to X-ray. Using elastic registration techniques based on the nuclear pattern of individual cells, we could describe the motion of individual RIF precisely within the nucleus. We show that DNA repair occurs in a limited number of large domains, within which multiple small RIFs form, merge and/or resolve with random motion following normal diffusion law. Large foci formation is shown to be mainly happening through the merging of smaller RIF rather than through growth of an individual focus. We estimate repair domain sizes of 7.5 to 11 µm 2 with a maximum number of ~15 domains per MCF10A cell. This work also highlights DDR which are specific to doses larger than 1 Gy such as rapid 53BP1 protein increase in the nucleus and foci diffusion rates that are significantly faster than for spontaneous foci movement. We hypothesize that RIF merging reflects a "stressed" DNA repair process that has been taken outside physiological conditions when too many DSB occur at once. High doses of ionizing radiation lead to RIF merging into repair domains which in turn increases DSB proximity and misrepair. Furthermore, such finding may therefore be critical to explain the supralinear dose dependence for chromosomal rearrangement and cell death measured after exposure to ionizing radiation.

59 BASIC BIOLOGICAL SCIENCES↗

Aftershock Identification Using a Paired Neural Network Applied to Constructed Data

This report is intended to detail the findings of our investigation of the applicability of machine learning to the task of aftershock identification. The ability to automatically identify nuisance aftershock events to reduce analyst workload when searching for events of interest is an important step in improving nuclear monitoring capabilities and while waveform cross - correlation methods have proven successful, they have limitations (e.g., difficulties with spike artifacts, multiple aftershocks in the same window) that machine learning may be able to overcome. Here we apply a Paired Neural Network (PNN) to a dataset consisting of real, high quality signals added to real seismic noises in order to work with controlled, labeled data and establish a baseline of the PNN's capability to identify aftershocks. We compare to waveform cross - correlation and find that the PNN performs well, outperforming waveform cross - correlation when classifying similar waveform pairs, i.e., aftershocks.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Robust automatic rigid registration of MRI and X-ray using external fiducial markers for XFM-guided interventional procedures

Purpose: In X-ray fused with MRI, previously gathered roadmap MRI volume images are overlaid on live X-ray fluoroscopy images to help guide the clinician during an interventional procedure. The incorporation of MRI data allows for the visualization of soft tissue that is poorly visualized under X-ray. The widespread clinical use of this technique will require fully automating as many components as possible. While previous use of this method has required time-consuming manual intervention to register the two modalities, in this article, the authors present a fully automatic rigid-body registration method. Methods: External fiducial markers that are visible under these two complimentary imaging modalities were used to register the X-ray images with the roadmap MR images. The method has three components: (a) The identification of the 3D locations of the markers from a full 3D MR volume, (b) the identification of the 3D locations of the markers from a small number of 2D X-ray fluoroscopy images, and (c) finding the rigid-body transformation that registers the two point sets in the two modalities. For part (a), the localization of the markers from MR data, the MR volume image was thresholded, connected voxels were segmented and labeled, and the centroids of the connected components were computed. For part (b), the X-ray projection images, produced by an image intensifier, were first corrected for distortions. Binary mask images of the markers were created from the distortion-corrected X-ray projection images by applying edge detection, pattern recognition, and image morphological operations. The markers were localized in the X-ray frame using an iterative backprojection-based method which segments voxels in the volume of interest, discards false positives based on the previously computed edge-detected projections, and calculates the locations of the true markers as the centroids of the clusters of voxels that remain. For part (c), a variant of the iterative closest point method was used to find correspondences between and register the two sets of points computed from MR and X-ray data. This knowledge of the correspondence between the two point sets was used to refine, first, the X-ray marker localization and then the total rigid-body registration between modalities. The rigid-body registration was used to overlay the roadmap MR image onto the X-ray fluoroscopy projections. Results: In 35 separate experiments, the markers were correctly registered to each other in 100% of the cases. When half the number of X-ray projections was used (10 X-ray projections instead of 20), the markers were correctly registered in all 35 experiments. The method was also successful in all 35 experiments when the number of markers was (retrospectively) halved (from 16 to 8). The target registration error was computed in a phantom experiment to be less than 2.4 mm. In two in vivo experiments, targets (interventional devices with pointlike metallic structures) inside the heart were successfully registered between the two modalities. Conclusions: The method presented can be used to automatically register a roadmap MR image to X-ray fluoroscopy using fiducial markers and as few as ten X-ray projections.

60 APPLIED LIFE SCIENCES↗

Using active learning to improve quasar identification for the DESI spectra processing pipeline

The Dark Energy Spectroscopic Instrument (DESI) survey uses an automatic spectral classification pipeline to classify spectra. QuasarNET is a convolutional neural network used as part of this pipeline originally trained using data from the Baryon Oscillation Spectroscopic Survey (BOSS). In this paper we implement an active learning algorithm to optimally select spectra to use for training a new version of the QuasarNET weights file using only DESI data, with the goal of improving classification accuracy. This active learning algorithm includes a novel outlier rejection step using a Self-Organizing Map to ensure we label spectra representative of the larger quasar sample observed in DESI. We perform two iterations of the active learning pipeline, assembling a final dataset of 5600 labeled spectra, a small subset of the approximately 1.3 million quasar targets in DESI's Data Release 1. When splitting the spectra into training and validation subsets we achieve similar performance to the previously trained weights file in completeness and purity calculated on the validation dataset but do so with less than one tenth of the amount of training data. The new weights also more consistently classify objects in the same way when used on unlabeled data compared to the old weights file. In the process of improving QuasarNET's classification accuracy we discovered a systemic error in QuasarNET's redshift estimation and used our findings to improve our understanding of QuasarNET's redshifts.

Machine learning↗

CLUSTERING-BASED FEATURE LEARNING ON VARIABLE STARS

The success of automatic classification of variable stars depends strongly on the lightcurve representation. Usually, lightcurves are represented as a vector of many descriptors designed by astronomers called features. These descriptors are expensive in terms of computing, require substantial research effort to develop, and do not guarantee a good classification. Today, lightcurve representation is not entirely automatic; algorithms must be designed and manually tuned up for every survey. The amounts of data that will be generated in the future mean astronomers must develop scalable and automated analysis pipelines. In this work we present a feature learning algorithm designed for variable objects. Our method works by extracting a large number of lightcurve subsequences from a given set, which are then clustered to find common local patterns in the time series. Representatives of these common patterns are then used to transform lightcurves of a labeled set into a new representation that can be used to train a classifier. The proposed algorithm learns the features from both labeled and unlabeled lightcurves, overcoming the bias using only labeled data. We test our method on data sets from the Massive Compact Halo Object survey and the Optical Gravitational Lensing Experiment; the results show that our classification performance is as good as and in some cases better than the performance achieved using traditional statistical features, while the computational cost is significantly lower. With these promising results, we believe that our method constitutes a significant step toward the automation of the lightcurve classification pipeline.

79 ASTRONOMY AND ASTROPHYSICS↗

Single-cell correlations of mRNA and protein content in a human monocytic cell line after LPS stimulation

The heterogeneity of mRNA and protein expression at the single-cell level can reveal fundamental information about cellular response to external stimuli, including the sensitivity, timing, and regulatory interactions of genes. Here we describe a fully automated system to digitally count the intron, mRNA, and protein content of up to five genes of interest simultaneously in single-cells. Full system automation of 3D microscope scans and custom image analysis routines allows hundreds of individual cells to be automatically segmented and the mRNA-protein content to be digitally counted. Single-molecule intron and mRNA content is measured by single-molecule fluorescence in-situ hybridization (smFISH), while protein content is quantified though the use of antibody probes. To mimic immune response to bacterial infection, human monocytic leukemia cells (THP-1) were stimulated with lipopolysaccharide (LPS), and the expression of two inflammatory genes, IL1β (interleukin 1β) and TNF-α (tumor necrosis factor α), were simultaneously quantified by monitoring the intron, mRNA, and protein levels over time. The simultaneous labeling of cellular content allowed for a series of correlations at the single-cell level to be explored, both in the progressive maturation of a single gene (intron-mRNA-protein) and comparative analysis between the two immune response genes. In the absence of LPS stimulation, mRNA expression of IL1β and TNF-α were uncorrelated. Following LPS stimulation, mRNA expression of the two genes became more correlated, consistent with a model in which IL1β and TNF-α upregulation occurs in parallel through independent mechanistic pathways. This smFISH methodology can be applied to different complex biological systems to provide valuable insight into highly dynamic gene mechanisms that determine cell plasticity and heterogeneity of cellular response.

59 BASIC BIOLOGICAL SCIENCES↗

PDF Entity Annotation Tool (PEAT)

While different text mining approaches – including the use of Artificial Intelligence (AI) and other machine based methods - continue to expand at a rapid pace, the tools used by researchers to create the labeled datasets required for training, modeling, and evaluation remain rudimentary. Labeled datasets contain the target attributes the machine is going to learn; for example, training an algorithm to delineate between images of a car or truck would generally require a set of images with a quantitative description of the underlying features of each vehicle type. Development of labeled textual data that can be used to build natural language machine learning models for scientific literature is not currently integrated into existing manual workflows used by domain experts. Published literature is rich with important information, such as different types of embedded text, plots, and tables that can all be used as inputs to train ML/natural language processing (NLP) models, when extracted and prepared in machine readable formats. Currently, both normalized data extraction of use to domain experts and extraction to support development of ML/NLP models are labor intensive and cumbersome manual processes. Automatic extraction of data and information from formats such as PDFs that are optimized for layout and human readability, not machine readability. The PDF (Portable Document Format) Entity Annotation Tool (PEAT) was developed with the goal of allowing users to annotate publications within their current print format, while also allowing those annotations to be captured in a machine-readable format. One of the main issues with traditional annotation tools is that they require transforming the PDF into plain text to facilitate the annotation process. While doing so lessens the technical challenges of annotating data, the user loses all structure and provenance that was inherent in the underlying PDF. Also, textual data extraction from PDFs can be an error prone process. Challenges include identifying sequential blocks of text and a multitude of document formats (multiple columns, font encodings, etc.). As a result of these challenges, using existing tools for development of NLP/ML models directly from PDFs is difficult because the generated outputs are not interoperable. We created a system that allows annotations to be completed on the original PDF document structure, with no plain text extraction. The result is an application that allows for easier and more accurate annotations. In addition, by including a feature that grants the user the ability to easily create a schema, we have developed a system that can be used to annotate text for different domain-centric schemas of relevance to subject matter experts. Different knowledge domains require distinct schemas and annotation tags to support machine learning.

97 MATHEMATICS AND COMPUTING↗