Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Recognition”

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 1,675 records · Page 93

The JCSG MR Pipeline: Optimized Alignments, Multiple Models And Parallel Searches

The success rate of molecular replacement (MR) falls considerably when search models share less than 35% sequence identity with their templates, but can be improved significantly by using fold-recognition methods combined with exhaustive MR searches. Models based on alignments calculated with fold-recognition algorithms are more accurate than models based on conventional alignment methods such as FASTA or BLAST, which are still widely used for MR. In addition, by designing MR pipelines that integrate phasing and automated refinement and allow parallel processing of such calculations, one can effectively increase the success rate of MR. Here, updated results from the JCSG MR pipeline are presented, which to date has solved 33 MR structures with less than 35% sequence identity to the closest homologue of known structure. By using difficult MR problems as examples, it is demonstrated that successful MR phasing is possible even in cases where the similarity between the model and the template can only be detected with fold-recognition algorithms. In the first step, several search models are built based on all homologues found in the PDB by fold-recognition algorithms. The models resulting from this process are used in parallel MR searches with different combinations of input parameters of the MR phasing algorithm. The putative solutions are subjected to rigid-body and restrained crystallographic refinement and ranked based on the final values of free R factor, figure of merit and deviations from ideal geometry. Finally, crystal packing and electron-density maps are checked to identify the correct solution. If this procedure does not yield a solution with interpretable electron-density maps, then even more alternative models are prepared. The structurally variable regions of a protein family are identified based on alignments of sequences and known structures from that family and appropriate trimmings of the models are proposed. All combinations of these trimmings are applied to the search models and the resulting set of models is used in the MR pipeline. It is estimated that with the improvements in model building and exhaustive parallel searches with existing phasing algorithms, MR can be successful for more than 50% of recognizable homologues of known structures below the threshold of 35% sequence identity. This implies that about one-third of the proteins in a typical bacterial proteome are potential MR targets.

59 BASIC BIOLOGICAL SCIENCES↗

A mathematical discussion concerning the performance of multilayer perceptron-type artificial neural networks through use of orthogonal bipolar vectors

This work proposes the use of orthogonal bipolar vectors (OBV) as targets for multilayer perceptron (MLP) Artificial neural networks pertaining to the pattern recognition field. Implications involving this proposal are supported by a proper mathematical discussion. This study shows that the use of OBV as targets improves the class identification performance of MLP networks for pattern recognition with more than two classes when compared to the widely used conventional target bipolar vector (CBV). In addition to the overall performance improvement, it is possible to obtain high recognition rates with fewer training epochs, while keeping a feasible generalization with other data. This behavior is related to the Euclidean distance of the points in the output space, which is much higher in the case of OBVs. Therefore, with a greater distance between the points in the output space, the MLP network becomes less prone to associate trained outputs with incorrect targets. The mathematical discussion in this study is based on the backpropagation training algorithm. In addition to the mathematical discussion, results from experiments with real data are presented. These results follow those obtained by mathematical derivations. Real data sets used in the experiments are available from: (a) the Semeion Handwritten Digit of Machine Learning Repository, international repository; (b) Iris Image Database from the Chinese Academy of Sciences—CASIA; (c) Australian Sign Language, signs of Machine Learning Repository, international repository. Experiments were conducted to measure the mean Euclidean distance between the MLP network output subjected to a given input pattern and outputs related to other pattern classifications. Experimental results show that MLP networks trained with OBVs were able to ensure a Euclidean distance between the output and incorrect targets greater than MLP networks trained with CBVs (by up to 4 times for handwritten, by up to 12 times for iris images and by up to 10 times for Auslan signs language). As such, the adoption of OBVs as target vectors for MLPs reduces the computational effort for training the network.

97 MATHEMATICS AND COMPUTING↗

Effects of non-invasive brain stimulation on associative memory

Associative memory refers to remembering the association between two items, such as a face and a name. It is a crucial part of daily life, but it is also one of the first aspects of memory performance that is impacted by aging and by Alzheimer’s disease. Evidence suggests that transcranial direct current stimulation (tDCS) can improve memory performance, but few tDCS studies have investigated its impact on associative memory. In addition, no prior study of the effects of tDCS on memory performance has systematically evaluated the impact of tDCS on different types of memory assessments, such as recognition and recall tests. In this study, we measured the effects of tDCS on associative memory performance in healthy adults, using both recognition and recall tests. Participants studied face-name pairs while receiving either active (30 minutes, 2 mA) or sham (30 minutes, 0.1 mA) stimulation with the anode placed at F9 and the cathode placed on the contralateral upper arm. Participants in the active stimulation group performed significantly better on the recall test than participants in the sham group, recalling 50% more names, on average, and making fewer recall errors. However, the two groups did not differ significantly in terms of their performance on the recognition memory test. This investigation provides evidence that stimulation at the time of study improves associative memory encoding, but that this memory benefit is evident only under certain retrieval conditions.

60 APPLIED LIFE SCIENCES↗

Particle track classification using quantum associative memory

Pattern recognition algorithms are commonly employed to simplify the challenging and necessary step of track reconstruction in sub-atomic physics experiments. Aiding in the discrimination of relevant interactions, pattern recognition seeks to accelerate track reconstruction by isolating signals of interest. In high collision rate experiments, such algorithms can be particularly crucial for determining whether to retain or discard information from a given interaction even before the data is transferred to tape. As data rates, detector resolution, noise, and inefficiencies increase, pattern recognition becomes more computationally challenging, motivating the development of higher efficiency algorithms and techniques. Quantum associative memory is an approach that seeks to exploits quantum mechanical phenomena to gain advantage in learning capacity, or the number of patterns that can be stored and accurately recalled. Here, we study quantum associative memory based on quantum annealing and apply it to the particle track classification. We focus on discrimination models based on Ising formulations of quantum associative memory model (QAMM) recall and quantum content-addressable memory (QCAM) recall. We characterize classification performance of these approaches as a function detector resolution, pattern library size, and detector inefficiencies, using the D-Wave 2000Q processor as a testbed. Discrimination criteria is set using both solution-state energy and classification labels embedded in solution states. We find that energy-based QAMM classification performs well in regimes of small pattern density and low detector inefficiency. In contrast, state-based QCAM achieves reasonably high accuracy recall for large pattern density and the greatest recall accuracy robustness to a variety of detector noise sources.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

A low-latency graph computer to identify metastable particles at the Large Hadron Collider for real-time analysis of potential dark matter signatures

Abstract Image recognition is a pervasive task in many information-processing environments. We present a solution to a difficult pattern recognition problem that lies at the heart of experimental particle physics. Future experiments with very high-intensity beams will produce a spray of thousands of particles in each beam-target or beam-beam collision. Recognizing the trajectories of these particles as they traverse layers of electronic sensors is a massive image recognition task that has never been accomplished in real time. We present a real-time processing solution that is implemented in a commercial field-programmable gate array using high-level synthesis. It is an unsupervised learning algorithm that uses techniques of graph computing. A prime application is the low-latency analysis of dark-matter signatures involving metastable charged particles that manifest as disappearing tracks.

47 OTHER INSTRUMENTATION↗

Grassmannian Diffusion Maps--Based Dimension Reduction and Classification for High-Dimensional Data

This work introduces the Grassmannian diffusion maps (GDMaps), a novel nonlinear dimensionality reduction technique that defines the affinity between points through their representation as low-dimensional subspaces corresponding to points on the Grassmann manifold. Here, the method is designed for applications, such as image recognition and data-based classification of constrained high-dimensional data where each data point itself is a high-dimensional object (i.e., a large matrix) that can be compactly represented in a lower-dimensional subspace. The GDMaps is composed of two stages. The first is a pointwise linear dimensionality reduction wherein each high-dimensional object is mapped onto the Grassmann manifold representing the low-dimensional subspace on which it resides. The second stage is a multipoint nonlinear kernel-based dimension reduction using diffusion maps to identify the subspace structure of the points on the Grassmann manifold. To this end, an appropriate Grassmannian kernel is used to construct the transition matrix of a random walk on a graph connecting points on the Grassmann manifold. Spectral analysis of the transition matrix yields low-dimensional Grassmannian diffusion coordinates embedding the data into a low-dimensional reproducing kernel Hilbert space. Further, a novel data classification/recognition technique is developed based on the construction of an overcomplete dictionary of reduced dimension whose atoms are given by the Grassmannian diffusion coordinates. Three examples are considered. First, a "toy" example shows that the GDMaps can identify an appropriate parametrization of structured points on the unit sphere. The second example demonstrates the ability of the GDMaps to revealing the intrinsic subspace structure of high-dimensional random field data. In the last ex- ample, a face recognition problem is solved considering face images subject to varying illumination conditions, changes in face expressions, and occurrence of occlusions. The technique presented high recognition rates (i.e., 95% in the best case) using a fraction of the data required by conventional methods.

42 ENGINEERING↗

Distillation Column Flooding Predictor

The Flooding Predictor™ is a patented advanced control technology proven in research at the Separations Research Program, University of Texas at Austin, to increase distillation column throughput by over 6%, while also increasing energy efficiency by 10%. The research was conducted under a U. S. Department of Energy Cooperative Agreement awarded to George Dzyacky of 2ndpoint, LLC. The Flooding Predictor™ works by detecting the incipient flood point and controlling the column closer to its actual hydraulic limit than historical practices have allowed. Further, the technology uses existing column instrumentation, meaning no additional refining infrastructure is required. Refiners often push distillation columns to maximize throughput, improve separation, or simply to achieve day-to-day optimization. Attempting to achieve such operating objectives is a tricky undertaking that can result in flooding. Operators and advanced control strategies alike rely on the conventional use of delta-pressure instrumentation to approximate the column’s approach to flood. But column delta-pressure is more an inference of the column’s approach to flood than it is an actual measurement of it. As a consequence, delta pressure limits are established conservatively in order to operate in a regime where the column is never expected to flood. As a result, there is much “left on the table” when operating in such a regime, i.e. the capacity difference between controlling the column to an upper delta-pressure limit and controlling it to the actual hydraulic limit. The Flooding Predictor™, an innovative pattern recognition technology, controls columns at their actual hydraulic limit, which research shows leads to a throughput increase of over 6%. Controlling closer to the hydraulic limit also permits operation in a sweet spot of increased energy-efficiency. In this region of increased column loading, the Flooding Predictor is able to exploit the benefits of higher liquid/vapor traffic that produce increased contact area and lead to substantial increases in separation efficiency – which translates to a 10% increase in energy efficiency on a BTU/bbl basis. The Flooding Predictor™ operates on the principle that between five to sixty minutes in advance of a flooding event, certain column variables experience an oscillation, a pre-flood pattern. The pattern recognition system of the Flooding Predictor™ utilizes the mathematical first derivative of certain column variables to identify the column’s pre-flood pattern(s). This pattern is a very brief, highly repeatable, simultaneous movement among the derivative values of certain column variables. While all column variables experience negligible random noise generated from the natural frequency of the process, subtle pre-flood patterns are revealed among sub-sets of the derivative values of column variables as the column approaches its hydraulic limit. The sub-set of column variables that comprise the pre-flood pattern is identified empirically through in a two-step process. First, 2ndpoint’s proprietary off-line analysis tool is used to mine historical data for pre-flood patterns. Second, the column is flood-tested to fine-tune the pattern recognition for commissioning. Then the Flooding Predictor™ is implemented as closed-loop advanced control strategy on the plant’s distributed control system (DCS), thus automating control of the column at its hydraulic limit.

02 PETROLEUM↗

Structural, kinetic, and thermodynamic studies of specificity designed HIV-1 protease

HIV-1 protease recognizes and cleaves more than 12 different substrates leading to viral maturation. While these substrates share no conserved motif, they are specifically selected for and cleaved by protease during viral life cycle. Drug resistant mutations evolve within the protease that compromise inhibitor binding but allow the continued recognition of all these substrates. While the substrate envelope defines a general shape for substrate recognition, successfully predicting the determinants of substrate binding specificity would provide additional insights into the mechanism of altered molecular recognition in resistant proteases. We designed a variant of HIV protease with altered specificity using positive computational design methods and validated the design using X-ray crystallography and enzyme biochemistry. The engineered variant, Pr3 (A28S/D30F/G48R), was designed to preferentially bind to one out of three of HIV protease's natural substrates; RT-RH over p2-NC and CA-p2. In kinetic assays, RT-RH binding specificity for Pr3 increased threefold compared to the wild-type (WT), which was further confirmed by isothermal titration calorimetry. Crystal structures of WT protease and the designed variant in complex with RT-RH, CA-p2, and p2-NC were determined. Structural analysis of the designed complexes revealed that one of the engineered substitutions (G48R) potentially stabilized heterogeneous flap conformations, thereby facilitating alternate modes of substrate binding. Our results demonstrate that while substrate specificity could be engineered in HIV protease, the structural pliability of protease restricted the propagation of interactions as predicted. These results offer new insights into the plasticity and structural determinants of substrate binding specificity of the HIV-1 protease.

59 BASIC BIOLOGICAL SCIENCES↗

Efficient multi-scale representation of visual objects using a biologically plausible spike-latency code and winner-take-all inhibition

Deep neural networks have surpassed human performance in key visual challenges such as object recognition, but require a large amount of energy, computation, and memory. In contrast, spiking neural networks (SNNs) have the potential to improve both the efficiency and biological plausibility of object recognition systems. Here we present a SNN model that uses spike-latency coding and winner-take-all inhibition (WTA-I) to efficiently represent visual stimuli using multi-scale parallel processing. Mimicking neuronal response properties in early visual cortex, images were preprocessed with three different spatial frequency (SF) channels, before they were fed to a layer of spiking neurons whose synaptic weights were updated using spike-timing-dependent-plasticity. We investigate how the quality of the represented objects changes under different SF bands and WTA-I schemes. We demonstrate that a network of 200 spiking neurons tuned to three SFs can efficiently represent objects with as little as 15 spikes per neuron. Furthermore, studying how core object recognition may be implemented using biologically plausible learning rules in SNNs may not only further our understanding of the brain, but also lead to novel and efficient artificial vision systems.

59 BASIC BIOLOGICAL SCIENCES↗

A Method to Recognize Anatomical Site and Image Acquisition View in X-ray Images

A method was developed to recognize anatomical site and image acquisition view automatically in 2D X-ray images that are used in image-guided radiation therapy. The purpose is to enable site and view dependent automation and optimization in the image processing tasks including 2D-2D image registration, 2D image contrast enhancement, and independent treatment site confirmation. The X-ray images for 180 patients of six disease sites (the brain, head-neck, breast, lung, abdomen, and pelvis) were included in this study with 30 patients each site and two images of orthogonal views each patient. A hierarchical multiclass recognition model was developed to recognize general site first and then specific site. Each node of the hierarchical model recognized the images using a feature extraction step based on principal component analysis followed by a binary classification step based on support vector machine. Given two images in known orthogonal views, the site recognition model achieved a 99% average F1 score across the six sites. If the views were unknown in the images, the average F1 score was 97%. If only one image was taken either with or without view information, the average F1 score was 94%. The accuracy of the site-specific view recognition models was 100%.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Characterization of the AT180 epitope of phosphorylated Tau protein by a combined nuclear magnetic resonance and fluorescence spectroscopy approach

Highlights: {yields} pThr231 of the Tau protein is necessary for the binding of the AT180 antibody. {yields} pSer235 of the Tau protein does not interfere with the AT180 recognition of pThr231. {yields} Epitope mapping is efficiently achieved by combining NMR and FRET spectroscopy. -- Abstract: We present here the characterization of the epitope recognized by the AT180 monoclonal antibody currently used to define an Alzheimer's disease (AD)-related pathological form of the phosphorylated Tau protein. Some ambiguity remains as to the exact phospho-residue(s) recognized by this monoclonal: pThr231 or both pThr231 and pSer235. To answer this question, we have used a combination of nuclear magnetic resonance (NMR) and fluorescence spectroscopy to characterize in a qualitative and quantitative manner the phospho-residue(s) essential for the epitope recognition. Data from the first step of NMR experiments are used to map the residues bound by the antibodies, which were found to be limited to a few residues. A fluorophore is then chemically attached to a cystein residue introduced close-by the mapped epitope, at arginine 221, by mutagenesis of the recombinant protein. The second step of Foerster resonance energy transfer (FRET) between the AT180 antibody tryptophanes and the phospho-Tau protein fluorophore allows to calculate a dissociation constant Kd of 30 nM. We show that the sole pThr231 is necessary for the AT180 recognition of phospho-Tau and that phosphorylation of Ser235 does not interfere with the binding.

60 APPLIED LIFE SCIENCES↗

Designer interface peptide grafts target estrogen receptor alpha dimerization

The nuclear transcription factor estrogen receptor alpha (ERα), triggered by its cognate ligand estrogen, regulates a variety of cellular signaling events. ERα is expressed in 70% of breast cancers and is a widely validated target for anti-breast cancer drug discovery. Administration of anti-estrogen to block estrogen receptor activation is still a viable anti-breast cancer treatment option but anti-estrogen resistance has been a significant bottle-neck. Dimerization of estrogen receptor is required for ER activation. Blocking ERα dimerization is therefore a complementary and alternative strategy to combat anti-estrogen resistance. Dimer interface peptide “I-box” derived from ER residues 503–518 specifically blocks ER dimerization. Recently using a comprehensive molecular simulation we studied the interaction dynamics of ERα LBDs in a homo-dimer. Based on this study, we identified three interface recognition peptide motifs LDKITDT (ERα residues 479–485), LQQQHQRLAQ (residues 497–506), and LSHIRHMSNK (residues 511–520) and reported the suitability of using LQQQHQRLAQ (ER 497–506) as a template to design inhibitors of ERα dimerization. Stability and self-aggregation of peptide based therapeutics poses a significant bottle-neck to proceed further. In this study utilizing peptide grafted to preserve their pharmacophoric recognition motif and assessed their stability and potential to block ERα mediated activity in silico and in vitro. The Grafted peptides blocked ERα mediated cell proliferation and viability of breast cancer cells but did not alter their apoptotic fate. We believe the structural clues identified in this study can be used to identify novel peptidometics and small molecules that specifically target ER dimer interface generating a new breed of anti-cancer agents. - Highlights: • Designer peptide grafts retain core molecular recognition motif during MD simulations. • Designer peptide grafts with Poly-ALA helix form stable complexes with estrogen receptor in silico. • Inhibitor peptides significantly decrease estrogen induced cell proliferation of ER positive breast cancer cells in vitro.

60 APPLIED LIFE SCIENCES↗

Characterization of the promoter region of bovine SIX4: Roles of E-box and MyoD in the regulation of basal transcription

Highlights: • Bovine SIX4 was highly expressed in the longissimus thoracis and in undifferentiated Qinchuan cattle muscle cells. • The core promoter of bovine SIX4 was located within the region −522/-193 relative to the TSS. • MyoD binding occurs in the core promoter of bovine SIX4 through MyoD and Ebox recognition sites and plays an important role in the transcriptional regulation of the bovine SIX4. The sine oculis homeobox 4 (SIX4) gene belongs to the Six gene family and encodes an evolutionarily conserved transcription factor. Previous studies have demonstrated that SIX4 plays an essential role in proper muscle regeneration. However, the mechanisms regulating SIX4 transcription remain elusive. In the present study, we determined that bovine SIX4 was highly expressed in the longissimus thoracis and in undifferentiated Qinchuan cattle muscle cells (QCMCs) and that its protein localizes to both the cytoplasm and the nucleus. To elucidate the bovine the molecular mechanisms of SIX4 regulation, 1.3 kb of the 5′-regulatory region was obtained. MyoD and Ebox recognition sites were identified in the core promoter region at −522/-193 of the bovine SIX4 using a series of 5′ deletion promoter plasmids in luciferase reporter assays. An electrophoretic mobility shift assay (EMSA) and chromatin immunoprecipitation (ChIP) assay in combination with site-directed mutation and siRNA interference demonstrated that MyoD binding occurs at MyoD and Ebox recognition sites through direct and indirect mechanisms and play important roles in the transcriptional regulation of the bovine SIX4 promoter. Taken together, these interactions provide insight into the regulatory mechanisms of SIX4 transcription in mediating skeletal muscle growth in cattle.

60 APPLIED LIFE SCIENCES↗

Structure of a [NiFe] hydrogenase maturation protease HycI provides insights into its substrate selectivity

Highlights: • TkHycI adopts a three-layer α/β/α fold with the conserved key residues. • TkHycI retains a larger loop than TkHybD, thereby creating a broad and deep cleft. • The sizable cleft of TkHycI is probably associated with its substrate recognition. The immature large subunit of [NiFe] hydrogenases undergoes C-terminal cleavage by a specific protease in the final step of the post-translational process before assembly with other subunits. It has been reported that the [NiFe] hydrogenase maturation protease HycI from Thermococcus kodakarensis (TkHycI) has the catalytic ability to target the membrane-bound hydrogenase large subunit MbhL from T. kodakarensis. However, the detailed mechanism of its substrate recognition remains elusive. We determined the crystal structure of TkHycI at 1.59 Å resolution to clarify how TkHycI recognizes its own substrate MbhL. Although the overall structure of TkHycI is similar to that of its homologous protease TkHybD, TkHycI adopts a larger loop than TkHybD, thereby creating a broad and deep cleft. We analyzed the structural properties of the TkHycI cleft probably involved in its substrate recognition. Our findings provide novel and profound insights into the substrate selectivity of TkHycI.

60 APPLIED LIFE SCIENCES↗

Specificity and utility of SubB2M, a new N-glycolylneuraminic acid lectin

Highlights: • SubB2M showed improved specificity and recognition for Neu5Gc containing glycans in a Neu5Ac/Neu5Gc glycan array and by surface plasmon resonance. • Far-western blot analysis showed that SubB2M preferentially binds to serum glycoproteins from bovine over human. • SubB2M detected Neu5Gc-glycoproteins spiked into normal human serum with greater sensitivity than SubB and anti-Neu5Gc IgY. • SubB2M will be a useful tool for the testing of serum and other bodily fluids for cancer diagnosis and prognosis. The B subunit of the subtilase cytotoxin (SubB) recognises N-glycolylneuraminic acid (Neu5Gc) containing glycans, the most prominent form of aberrant glycosylation in human cancers. We have previously engineered SubB by construction of a SubB{sub ΔS106/ΔT107} mutant (SubB2M) for greater specificity and enhanced recognition of Neu5Gc containing glycans. In this study, we further explore the utility of SubB2M as a Neu5Gc lectin by showing its improved specificity and recognition for Neu5Gc containing glycans over the wild-type SubB protein and an anti-Neu5Gc IgY antibody in a N-acetylneuraminic acid (Neu5Ac)/Neu5Gc glycan array and by surface plasmon resonance. Far-western blot analysis showed that SubB2M preferentially binds to bovine serum glycoproteins over human serum glycoproteins. SubB2M was also able to detect Neu5Gc containing bovine glycoproteins spiked into normal human serum with greater sensitivity than the wild-type SubB and the anti-Neu5Gc IgY antibody. These results suggest that SubB2M will be a useful tool for the testing of serum and other bodily fluids for cancer diagnosis and prognosis.

60 APPLIED LIFE SCIENCES↗

Structures of Human Antibodies Bound to SARS-CoV-2 Spike Reveal Common Epitopes and Recurrent Features of Antibodies

Neutralizing antibody responses to coronaviruses mainly target the receptor-binding domain (RBD) of the trimeric spike. Here, we characterized polyclonal immunoglobulin Gs (IgGs) and Fabs from COVID-19 convalescent individuals for recognition of coronavirus spikes. Plasma IgGs differed in their focus on RBD epitopes, recognition of alpha- and beta-coronaviruses, and contributions of avidity to increased binding/ neutralization of IgGs over Fabs. Using electron microscopy, we examined specificities of polyclonal plasma Fabs, revealing recognition of both S1 A and RBD epitopes on SARS-CoV-2 spike. Moreover, a 3.4 Å cryoelectron microscopy (cryo-EM) structure of a neutralizing monoclonal Fab-spike complex revealed an epitope that blocks ACE2 receptor binding. Modeling based on these structures suggested different potentials for inter-spike crosslinking by IgGs on viruses, and characterized IgGs would not be affected by identified SARS-CoV-2 spike mutations. Overall, our studies structurally define a recurrent anti-SARS-CoV-2 antibody class derived from VH3-53/VH3-66 and similarity to a SARS-CoV VH3-30 antibody, providing criteria for evaluating vaccine-elicited antibodies.

59 BASIC BIOLOGICAL SCIENCES↗

Using dorsal surface for individual identification of dairy calves through 3D deep learning algorithms

Advances in machine learning techniques have allowed the development of computer vision systems (CVS) that can accurately predict several phenotypes of interest for livestock operations. In this context, 3D images taken from a top-down view are particularly useful for estimating body condition score, growth development, and body biometrics in cattle. Frequently, such CVS rely on identification (ID) systems, such as electronic tags, as a way to match animal ID and the predicted phenotype. However, the same 3D images used to predict body weight and other animal biometrics could be adopted for animal recognition as well. Such alternative would optimize CVS to recognize animal ID and monitor growth development simultaneously while leveraging the same hardware infrastructure. Furthermore, this strategy could be used to recognize animals with similar color patterns. Nonetheless, growing animals are continuously changing body shape, which could limit its use as an invariant feature for pattern recognition. Thus, the objectives of this study were: (1) to compare algorithms for different 3D object representations to identify individual animals; and (2) to evaluate how short-term changes in body shape due to animal growth affect the predictive performance of these algorithms. For objective 1, the algorithms were trained (n = 4,558) and tested (n = 1,139) using images from 38 Holstein calves. For objective 2, we designed three different experiments using images (n = 2,347) from five Holstein calves taken over six weeks during their growing period, always training and testing on different weeks. Each experiment evaluated how changing a different parameter of the image capturing procedure affected the predictive ability of the trained algorithms. In the first experiment, we varied the total number of images per animal in the training set; in the second experiment, we varied the number of weeks while keeping a fixed number of images in the training set; and in the third experiment, we skipped weeks between images in the training and test sets. The F 1 score for objective (1) was up to 0.804 when testing with the last frames of each video, and up to 0.959 when using random frames for testing. For objective (2), the F 1 score was up to 0.947 for the first experiment when using 130 images per animal; up to 0.979 for the second experiment when using all five weeks; and up to 0.917 when not skipping weeks between training and testing. In conclusion, these results show that deep learning algorithms can be used to identify individual animals through their dorsal area 3D surfaces, and, from our experiments using calves in their growing period, that they are robust enough to account for changes in body shape and size, making them a promising tool for animal recognition during growth.

3D neural networks↗

Machine Learning Approaches for Predicting Radiation Therapy Outcomes: A Clinician's Perspective

Radiation oncology has always been deeply rooted in modeling, from the early days of isoeffect curves to the contemporary Quantitative Analysis of Normal Tissue Effects in the Clinic (QUANTEC) initiative. In recent years, medical modeling for both prognostic and therapeutic purposes has exploded thanks to increasing availability of electronic data and genomics. One promising direction that medical modeling is moving toward is adopting the same machine learning methods used by companies such as Google and Facebook to combat disease. Broadly defined, machine learning is a branch of computer science that deals with making predictions from complex data through statistical models. These methods serve to uncover patterns in data and are actively used in areas such as speech recognition, handwriting recognition, face recognition, “spam” filtering (junk email), and targeted advertising. Although multiple radiation oncology research groups have shown the value of applied machine learning (ML), clinical adoption has been slow due to the high barrier to understanding these complex models by clinicians. Here, we present a review of the use of ML to predict radiation therapy outcomes from the clinician's point of view with the hope that it lowers the “barrier to entry” for those without formal training in ML. We begin by describing 7 principles that one should consider when evaluating (or creating) an ML model in radiation oncology. We next introduce 3 popular ML methods—logistic regression (LR), support vector machine (SVM), and artificial neural network (ANN)—and critique 3 seminal papers in the context of these principles. Although current studies are in exploratory stages, the overall methodology has progressively matured, and the field is ready for larger-scale further investigation.

62 RADIOLOGY AND NUCLEAR MEDICINE↗