Examinando por Autor "Viatkin, Dimitri"

Mostrando 1 - 3 de 3
  • Miniatura
    Ítem
    Advanced artificial intelligence methods applied to societal challenges in biomedical engineering
    (Universidad de Deusto, 2023-01-20) Viatkin, Dimitri; García-Zapirain, Begoña; Méndez Zorrilla, Amaia; Angulo Martínez, Ignacio; Facultad de Ingeniería; Programa de Doctorado en Ingeniería para la Sociedad de la Información y Desarrollo Sostenible por la Universidad de Deusto
    In recent years, artificial intelligence and machine learning algorithms have been increasingly used not only in scientific, but also in applied and societal fields. This is due to the development of computing power, technological development and the development of the algorithms used. Developed algorithms of artificial intelligence begin to be introduced also in medicine, engineering, bioengineering, biomedical and other frontier areas, where different knowledge areas touch each other. Development and training of artificial intelligence algorithms for solving problems, which are at the interaction boundary of different fields of science, can improve the quality of interaction between experts from different fields, expand the frontiers of knowledge and solve applied problems in the areas under study. This dissertation examines the possibilities of using artificial intelligence methods in biomedical engineering tasks, at the edge of medicine and engineering. The possibilities of analysis, development, use and interpretation of artificial intelligence algorithms in applied problems for sustainable development of society, medical and industrial development are considered. The dissertation consists of two case studies conducted in Spain and Russia, each using a different methodology and approach to analysis. The first case study explores the application of deep learning to the task of measuring the position of patients' fingers in multiple sclerosis. Tracking the limited degree of mobility of the fingers on the hand can be used as a marker to characterize the course of multiple sclerosis and the success of the treatment prescribed. The objectives of this case study were to review and analyze the literature on the various methods available for assessing finger position, to collect and prepare data for a single camera-based computer vision system designed to detect finger position, and to train and test a neural network based on a neural network for assessing finger position. The second case study explores the potential of deep learning methods for materials analysis and the possibility of applying them for biomedical purposes. This case study explores the potential of neural networks to analyze the properties and structure of materials with different amounts of data and different representations. The generation of materials based on a number of incomplete parameters with limited data has been studied. Algorithms for processing different types of material data representations and their parameters have been studied. In this case study, the following tasks were accomplished: literature review and analysis on various available material analysis methods, collectiovn and preparation of data for a material analysis system with different structures and parameters, training and testing the neural network on the collected data. The neural network prediction of critical superconductivity temperature for materials based on their chemical formula was considered. The prediction of the reduced glass transition temperature of metal alloys based on a neural network was considered. The prediction of material composition based on the required physical parameters for cellulosic materials was considered. The use of generative-adversarial networks to predict the properties and composition of metal alloys based on incomplete material information with an acceptable range of predicted parameters was considered. The second case study demonstrates the development of the idea of applying neural networks to materials problems, from predicting a single parameter from a chemical formula, to predicting physical parameters and material composition based on incomplete data.
  • Miniatura
    Ítem
    Deep learning approach for prediction of critical temperature of superconductor materials described by chemical formulas
    (Frontiers Media S.A., 2021-10-27) Viatkin, Dimitri ; García-Zapirain, Begoña ; Méndez Zorrilla, Amaia ; Zakharov, Maxim
    This paper proposes a novel neural network architecture and its ensembles to predict the critical superconductivity temperature of materials based on their chemical formula. The research describes the methods and processes of extracting data from the chemical formula and preparing these extracted data for use in neural network training using TensorFlow. In our approach, recurrent neural networks are used including long short-term memory layers and neural networks based on one-dimensional convolution layers for data analysis. The proposed model is an ensemble of pre-trained neural network architectures for the prediction of the critical temperature of superconductors based on their chemical formula. The architecture of seven pre-trained neural networks is based on the long short-term memory layers and convolution layers. In the final ensemble, six neural networks are used: one network based on LSTM and four based on convolutional neural networks, and one embedding ensemble of convolution neural networks. LSTM neural network and convolution neural network were trained in 300 epochs. Ensembles of models were trained in 20 epochs. All neural networks are trained in two stages. At both stages, the optimizer Adam was used. In the first stage, training was carried out by the function of losses Mean Absolute Error (MAE) with the value of optimizer learning rate equal to 0.001. In the second stage, the previously trained model was trained by the function of losses Mean Squared Error (MSE) with a learning rate equal to 0.0001. The final ensemble is trained with a learning rate equal to 0.00001. The final ensemble model has the following accuracy values: MAE is 4.068, MSE is 67.272, and the coefficient of determination (R2) is 0.923. The final model can predict the critical temperature for the chemistry formula with an accuracy of 4.068°.
  • Miniatura
    Ítem
    Deep learning techniques applied to predict and measure finger movement in patients with multiple sclerosis
    (MDPI AG, 2021-04-01) Viatkin, Dimitri ; García-Zapirain, Begoña ; Méndez Zorrilla, Amaia
    This research focuses on the development of a system for measuring finger joint angles based on camera image and is intended for work within the field of medicine to track the movement and limits of hand mobility in multiple sclerosis. Measuring changes in hand mobility allows the progress of the disease and its treatment process to be monitored. A static RGB camera without depth vision was used in the system developed, with the system receiving only the image from the camera and no other input data. The research focuses on the analysis of each image in the video stream independently of other images from that stream, and 12 measured hand parameters were chosen as follows: 3 joint angles for the index finger, 3 joint angles for the middle finger, 3 joint angles for the ring finger, and 3 joint angles for the pinky finger. Convolutional neural networks were used to analyze the information received from the camera, and the research considers neural networks based on different architectures and their combinations as follows: VGG16, MobileNet, MobileNetV2, InceptionV3, DenseNet, ResNet, and convolutional pose machine. The final neural network used for image analysis was a modernized neural network based on MobileNetV2, which obtained the best mean absolute error value of 4.757 degrees. Additionally, the mean square error was 67.279 and the root mean square error was 8.202 degrees. This neural network analyzed a single image from the camera without using other sensors. For its part, the input image had a resolution of 512 by 512 pixels, and was processed by the neural network in 7-15 milliseconds by GPU Nvidia 2080ti. The resulting neural network developed can measure finger joint angle values for a hand with non-standard parameters and positions.