Pasar al contenido principal

MONITORIZACION AVANZADA DE COMPORTAMIENTOS DE CAENORHABDITIS ELEGANS, BASADA EN VISION ACTIVA, PARA ANALIZAR FUNCION COGNITIVA Y ENVEJECIMIENTO

PRE2019-088214

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal de Promoción del Talento y su Empleabilidad en I+D+i
Subprograma Subprograma Estatal de Formación
Convocatoria Contratos Predoctorales
Año convocatoria 2019
Plan Estatal 2017-2020 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSITAT POLITÈCNICA DE VALÈNCIA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 5
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Tesis doctoral (DoctoralThesis)

Desarrollo de técnicas avanzadas de seguimiento de posturas para reconocimiento de comportamientos de C. elegans

  • Layana Castro, Pablo Emmanuel
[ES] El objetivo principal de esta tesis es el desarrollo de técnicas avanzadas de seguimiento de posturas para reconocimiento de comportamientos del Caenorhabditis elegans o C. elegans. El C. elegans es una clase de nematodo utilizado como organismo modelo para el estudio y tratamientos de diferentes enfermedades patológicas así como neurodegenerativas. Su comportamiento ofrece información valiosa para la investigación de nuevos fármacos (o productos alimenticios y cosméticos saludables) en el estudio de lifespan y healthspan. Al día de hoy, muchos de los ensayos con C. elegans se realizan de forma manual, es decir, usando microscopios para seguirlos y observar sus comportamientos o en laboratorios más modernos utilizando programas específicos. Estos programas no son totalmente automáticos, requieren ajuste de parámetros. Y en otros casos, son programas para visualización de imágenes donde el operador debe etiquetar maualmente el comportamiento de cada C. elegans. Todo esto se traduce a muchas horas de trabajo, lo cual se puede automatizar utilizando técnicas de visión por computador. Además de poder estimar indicadores de movilidad con mayor precisión que un operador humano. El problema principal en el seguimiento de posturas de C. elegans en placas de Petri son las agregaciones entre nematodos o con ruido del entorno. La pérdida o cambios de identidad son muy comunes ya sea de forma manual o usando programas automáticos/semi-automáticos. Y este problema se vuelve más complicado aún en imágenes de baja resolución. Los programas que automatizan estas tareas de seguimiento de posturas trabajan con técnicas de visión por computador usando técnicas tradicionales de procesamiento de imágenes o técnicas de aprendizaje profundo. Ambas técnicas han demostrado excelentes resultados en la detección y seguimiento de posturas de C. elegan}. Por un lado, técnicas tradicionales utilizan algoritmos/optimizadores para obtener la mejor solución, mientras que las técnicas de aprendizaje profundo aprenden de forma automática características del conjunto de datos de entrenamiento. El problema con las técnicas de aprendizaje profundo es que necesitan un conjunto de datos dedicado y grande para entrenar los modelos. La metodología utilizada para el desarrollo de esta tesis (técnicas avanzadas de seguimiento de posturas) se encuadran dentro del área de investigación de la visión artificial. Y ha sido abordada explorando ambas ramas de visión por computador para resolver los problemas de seguimiento de posturas de C. elegans en imágenes de baja resolución. La primera parte, es decir, secciones 1 y 2, capítulo 2, utilizó técnicas tradicionales de procesamiento de imágenes para realizar la detección y seguimiento de posturas de los C. elegans. Para ello se propuso una nueva técnica de esqueletización y dos nuevos criterios de evaluación para obtener mejores resultados de seguimiento, detección, y segmentación de posturas. Las siguientes secciones del capítulo 2 utilizan técnicas de aprendizaje profundo, y simulación de imágenes sintéticas para entrenar modelos y mejorar los resultados de detección y predicción de posturas. Los resultados demostraron ser más rápidos y más precisos en comparación con técnicas tradicionales. También se demostró que los métodos de aprendizaje profundo son más robustos ante la presencia de ruido en la placa., [CA] L'objectiu principal d'aquesta tesi és el desenvolupament de tècniques avançades de seguiment de postures per a reconeixement de comportaments del Caenorhabditis elegans o C. elegans. El C. elegans és una classe de nematodo utilitzat com a organisme model per a l'estudi i tractaments de diferents malalties patològiques així com neurodegeneratives. El seu comportament ofereix informació valuosa per a la investigació de nous fàrmacs (o productes alimentosos i cosmètics saludables) en l'estudi de lifespan i healthspan. Al dia de hui, molts dels assajos amb C. elegans es realitzen de manera manual, és a dir, usant microscopis per a seguir-los i observar els seus comportaments o en laboratoris més moderns utilitzant programes específics. Aquests programes no són totalment automàtics, requereixen ajust de paràmetres. I en altres casos, són programes per a visualització d'imatges on l'operador ha d'etiquetar maualment el comportament de cada C. elegans. Tot això es tradueix a moltes hores de treball, la qual cosa es pot automatitzar utilitzant tècniques de visió per computador. A més de poder estimar indicadors de mobilitat amb major precisió que un operador humà. El problema principal en el seguiment de postures de C. elegans en plaques de Petri són les agregacions entre nematodes o amb soroll de l'entorn. La pèrdua o canvis d'identitat són molt comuns ja siga de manera manual o usant programes automàtics/semi-automàtics. I aquest problema es torna més complicat encara en imatges de baixa resolució. Els programes que automatitzen aquestes tasques de seguiment de postures treballen amb tècniques de visió per computador usant tècniques tradicionals de processament d'imatges o tècniques d'aprenentatge profund. Totes dues tècniques han demostrat excel·lents resultats en la detecció i seguiment de postures de C. elegans. D'una banda, tècniques tradicionals utilitzen algorismes/optimizadors per a obtindre la millor solució, mentre que les tècniques d'aprenentatge profund aprenen de manera automàtica característiques del conjunt de dades d'entrenament. El problema amb les tècniques d'aprenentatge profund és que necessiten un conjunt de dades dedicat i gran per a entrenar els models. La metodologia utilitzada per al desenvolupament d'aquesta tesi (tècniques avançades de seguiment de postures) s'enquadren dins de l'àrea d'investigació de la visió artificial. I ha sigut abordada explorant totes dues branques de visió per computador per a resoldre els problemes de seguiment de postures de C. elegans en imatges de baixa resolució. La primera part, és a dir, secció 1 i 2, capítol 2, va utilitzar tècniques tradicionals de processament d'imatges per a realitzar la detecció i seguiment de postures dels C. elegans. Per a això es va proposar una nova tècnica de esqueletizació i dos nous criteris d'avaluació per a obtindre millors resultats de seguiment, detecció i segmentació de postures. Les següents seccions del capítol 2 utilitzen tècniques d'aprenentatge profund i simulació d'imatges sintètiques per a entrenar models i millorar els resultats de detecció i predicció de postures. Els resultats van demostrar ser més ràpids i més precisos en comparació amb tècniques tradicionals. També es va demostrar que els mètodes d'aprenentatge profund són més robustos davant la presència de soroll en la placa., [EN] The main objective of this thesis is the development of advanced posture-tracking techniques for behavioural recognition of Caenorhabditis elegans or C. elegans. C. elegans is a kind of nematode used as a model organism for the study and treatment of different pathological and neurodegenerative diseases. Their behaviour provides valuable information for the research of new drugs (or healthy food and cosmetic products) in the study of lifespan and healthspan. Today, many of the tests on C. elegans are performed manually, i.e. using microscopes to track them and observe their behaviour, or in more modern laboratories using specific software. These programmes are not fully automatic, requiring parameter adjustment. And in other cases, they are programmes for image visualisation where the operator must label the behaviour of each C. elegans manually. All this translates into many hours of work, which can be automated using computer vision techniques. In addition to being able to estimate mobility indicators more accurately than a human operator. The main problem in tracking C. elegans postures in Petri dishes is aggregations between nematodes or with noise from the environment. Loss or changes of identity are very common either manually or using automatic/semi-automatic programs. And this problem becomes even more complicated in low-resolution images. Programs that automate these pose-tracking tasks work with computer vision techniques using either traditional image processing techniques or deep learning techniques. Both techniques have shown excellent results in the detection and tracking of C. elegans postures. On the one hand, traditional techniques use algorithms/optimizers to obtain the best solution, while deep learning techniques automatically learn features from the training dataset. The problem with deep learning techniques is that they need a dedicated and large dataset to train the models. The methodology used for the development of this thesis (advanced posture-tracking techniques) falls within the research area of computer vision. It has been approached by exploring both branches of computer vision to solve the posture-tracking problems of C. elegans in low-resolution images. The first part, i.e. sections 1 and 2, chapter 2, used traditional image processing techniques to perform posture detection and tracking of C. elegans. For this purpose, a new skeletonization technique and two new evaluation criteria were proposed to obtain better posture-tracking, detection, and segmentation results. The next sections of chapter 2 use deep learning techniques, and synthetic image simulation to train models and improve posture detection and prediction results. The results proved to be faster and more accurate compared to traditional techniques. Deep learning methods were also shown to be more robust in the presence of plate noise., This research was supported by Ministerio de Ciencia, Innovación y Universidades [RTI2018-094312-B-I00 (European FEDER funds); FPI PRE2019-088214], and also was supported by Universitat Politècnica de València [“Funding for open access charge: Uni- versitat Politècnica de València”]. The author received a scholarship from the grant: Ayudas para contratos predoctorales para la formación de doctores 2019.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Skeletonizing Caenorhabditis elegans Based on U-Net Architectures Trained with a Multi-worm Low-Resolution Synthetic Dataset

  • Layana-Castro, Pablo Emmanuel
  • Navarro Moya, Francisco
  • García-Garví, Antonio|||0000-0002-3676-8287
  • Sánchez Salmerón, Antonio José|||0000-0003-1896-5356
[EN] Skeletonization algorithms are used as basic methods to solve tracking problems, pose estimation, or predict animal group behavior. Traditional skeletonization techniques, based on image processing algorithms, are very sensitive to the shapes of the connected components in the initial segmented image, especially when these are low-resolution images. Currently, neural networks are an alternative providing more robust results in the presence of image-based noise. However, training a deep neural network requires a very large and balanced dataset, which is sometimes too expensive or impossible to obtain. This work proposes a new training method based on a custom-generated dataset with a synthetic image simulator. This training method was applied to different U-Net neural networks architectures to solve the problem of skeletonization using low-resolution images of multiple Caenorhabditis elegans contained in Petri dishes measuring 55 mm in diameter. These U-Net models had only been trained and validated with a synthetic image; however, they were successfully tested with a dataset of real images. All the U-Net models presented a good generalization of the real dataset, endorsing the proposed learning method, and also gave good skeletonization results in the presence of image-based noise. The best U-Net model presented a significant improvement of 3.32% with respect to previous work using traditional image processing techniques., ADM Nutrition, Biopolis S.L. and Archer Daniels Midland supplied the C. elegans plates. Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). Mrs. Maria-Gabriela Salazar-Secada developed the skeleton annotation application. Mr. Jordi Tortosa-Grau and Mr. Ernesto-Jesus Rico-Guardioa annotated worm skeletons.Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature. This study was supported by the Plan Nacional de I+D with Project RTI2018-094312-B-I00, FPI Predoctoral contract PRE2019-088214 and by European FEDER funds.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Towards generalization for Caenorhabditis elegans detection

  • Escobar-Benavides, Santiago Nahuel
  • García-Garví, Antonio|||0000-0002-3676-8287
  • Sánchez Salmerón, Antonio José|||0000-0003-1896-5356
  • Layana-Castro, Pablo Emmanuel
[EN] The nematode Caenorhabditis elegans (C. elegans) is of significant interest for research into neurodegenerative diseases, aging, and drug screening. However, conducting these assays manually is a tedious and time-consuming process. This paper proposes a methodology to achieve a generalist C. elegans detection algorithm, as previous work only focused on dataset-specific detection, tailored exclusively to the characteristics and appearance of the images in a given dataset. The main aim of our study is to achieve a solution that allows for robust detection, regardless of the image-capture system used, with the potential to serve as a basis for the automation of numerous assays. These potential applications include worm counting, worm tracking, motion detection and motion characterization. To train this model, a dataset consisting of a wide variety of appearances adopted by C. elegans has been curated and dataset augmentation methods have been proposed and evaluated, including synthetic image generation. The results show that the model achieves an average precision of 89.5% for a wide variety of C. elegans appearances that were not used during training, thereby validating its generalization capabilities., This study was supported by Universidad Politecnica de Valencia through Instituto de Automatica e Informatica Industrial, FPI Predoctoral contract PRE2019-088214, Ministerio de Universidades (Spain) under grant FPU20/02639 and by European FEDER funds. The authors also thank the EU-FEDER Comunitat Valenciana 2014-2020 grant IDIFEDER/2018/025. ADM Nutrition, Biopolis SL, and Archer Daniels Midland provided support in the supply of C. ele-gans.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Automatic segmentation of Caenorhabditis elegans skeletons in worm aggregations using improved U-Net in low-resolution image sequences

  • Layana-Castro, Pablo Emmanuel
  • García-Garví, Antonio|||0000-0002-3676-8287
  • Sánchez Salmerón, Antonio José|||0000-0003-1896-5356
[EN] Pose estimation of C. elegans in image sequences is challenging and even more difficult in low-resolution images. Problems range from occlusions, loss of worm identity, and overlaps to aggregations that are too complex or difficult to resolve, even for the human eye. Neural networks, on the other hand, have shown good results in both low-resolution and high-resolution images. However, training in a neural network model requires a very large and balanced dataset, which is sometimes impossible or too expensive to obtain. In this article, a novel method for predicting C. elegans poses in cases of multi-worm aggregation and aggregation with noise is proposed. To solve this problem we use an improved U-Net model capable of obtaining images of the next aggregated worm posture. This neural network model was trained/validated using a custom-generated dataset with a synthetic image simulator. Subsequently, tested with a dataset of real images. The results obtained were greater than 75% in precision and 0.65 with Intersection over Union (IoU) values., Prof. Antonio-Jose Sanchez-Salmeron; Pablo E. Layana Castro; Antonio Garcia Garvi were supported by Ministerio de Ciencia, Innovacion y Universidades [RTI2018-094312-B-I00 (European FEDER funds); FPI PRE2019-088214; FPU20/02639]. This work was supported by Universitat Politecnica de Valencia ["Funding for open access charge: Universitat Politecnica de Valencia"].
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Analysis of a C. elegans lifespan prediction method based on a bimodal neural network and uncertainty estimation

  • García-Garví, Antonio|||0000-0002-3676-8287
  • Sánchez Salmerón, Antonio José|||0000-0003-1896-5356
  • Layana-Castro, Pablo Emmanuel
[EN] In recent decades, assays with the nematode Caenorhabditis elegans (C. elegans) have enabled great advances to be made in research on aging. However, performing these assays manually is a laborious task. To solve this problem, numerous C. elegans assay automation techniques are being developed to increase throughput and accuracy. In this paper, a method for predicting the lifespan of C. elegans nematodes using a bimodal neural network is proposed and analyzed. Specifically, the model uses the sequence of images and the count of live C. elegans up to the current day to predict the lifespan curve termination. This network has been trained using a simulator to avoid the labeling costs of training such a model. In addition, a method for estimating the uncertainty of the model predictions has been proposed. Using this uncertainty, a criterion has been analyzed to decide at what point the assay could be halted and the user could rely on the model¿s predictions. The method has been analyzed and validated using real experiments. The results show that uncertainty is reduced from the mean lifespan and that most of the predictions obtained do not present statistically significant differences with respect to the curves obtained manually., This study was supported by the Plan Nacional de I+D with Project RTI2018-094312-B-I00, European FEDER funds and by Ministerio de Universidades (Spain) under Grant FPU20/02639 and Grant PRE2019-088214. Funding for open access charge: Universitat Politecnica de Valencia. ADM Nutrition, Biopolis SL, and Archer Daniels Midland provided support in the supply of C. elegans.