Ultrasound tomography enhancement by signal feature extraction with modular machine learning method.

Robust and reliable diagnostic methods are desired in various types of industries. This article presents a novel approach to object detection in industrial or general ultrasound tomography. The key idea is to analyze the time-dependent ultrasonic signal recorded by three independent transducers of an experimental system. It focuses on finding common or related characteristics of these signals using custom-designed deep neural network models. In principle, models use convolution layers to extract common features of signals, which are passed to dense layers responsible for predicting the number of objects or their locations and sizes. Predicting the number and properties of objects are characterized by a high value of the coefficient of determination R2 = 99.8% and R2 = 98.4%, respectively. The proposed solution can result in a reliable and low-cost method of object detection for various industry sectors.

Autonomous Face Classification Online Self-Training System Using Pretrained ResNet50 and Multinomial Naïve Bayes.

This paper presents a novel, autonomous learning system working in real-time for face recognition. Multiple convolutional neural networks for face recognition tasks are available; however, these networks need training data and a relatively long training process as the training speed depends on hardware characteristics. Pretrained convolutional neural networks could be useful for encoding face images (after classifier layers are removed). This system uses a pretrained ResNet50 model to encode face images from a camera and the Multinomial Naïve Bayes for autonomous training in the real-time classification of persons. Faces of several persons visible in a camera are tracked using special cognitive tracking agents who deal with machine learning models. After a face in a new position of the frame appears (in a place where there was no face in the previous frames), the system checks if it is novel or not using a novelty detection algorithm based on an SVM classifier; if it is unknown, the system automatically starts training. As a result of the conducted experiments, one can conclude that good conditions provide assurance that the system can learn the faces of a new person who appears in the frame correctly. Based on our research, we can conclude that the critical element of this system working is the novelty detection algorithm. If false novelty detection works, the system can assign two or more different identities or classify a new person into one of the existing groups.

A Tactile Skin System for Touch Sensing with Ultrasound Tomography.

The tomographic imaging method is promising in large-area touch-sensing applications. This paper presents a new type of such touch sensor using ultrasonic tomography (UST) via sound attenuation imaging. UST is gaining popularity as a portable, fast, and inexpensive imaging system for medical and industrial applications. UST can be developed in different operation modes. A transmission mode UST is being investigated as a force- and touch-sensitive skin. A prototype skin sensor was developed in a 200 mm diameter circular UST array containing two sets of 16 transducers, with one operating at a central frequency of 40 kHz and the other at 300 kHz. The extension of the sensor in terms of dimension, up to 400 mm diameter, and number of sensors, up to 32 transducers, is possible where eight points of contact were reconstructed successfully. The medium contains a 20 mm high water region, and a soft silicone membrane covers the liquid region. When touchpoints or forces are applied to the soft skin of the membrane, the sound pathway is disrupted, resulting in an image of the touch position and touch force intensity using a tomographic UST algorithm. Several static and dynamic experiments are conducted to demonstrate this novel application of UST. In addition, a correlation analysis is carried out to establish the force quantification potential for the UST-based tactile skin.

Application of Machine Learning Algorithms to the Discretization Problem in Wearable Electrical Tomography Imaging for Bladder Tracking.

The article presents the implementation of artificial intelligence algorithms for the problem of discretization in Electrical Impedance Tomography (EIT) adapted for urinary tract monitoring. The primary objective of discretization is to create a finite element mesh (FEM) classifier that will separate the inclusion elements from the background. In general, the classifier is designed to detect the area of elements belonging to an inclusion revealing the shape of that object. We show the adaptation of supervised learning methods such as logistic regression, decision trees, linear and quadratic discriminant analysis to the problem of tracking the urinary bladder using EIT. Our study focuses on developing and comparing various algorithms for discretization, which perfectly supplement methods for an inverse problem. The innovation of the presented solutions lies in the originally adapted algorithms for EIT allowing for the tracking of the bladder. We claim that a robust measurement solution with sensors and statistical methods can track the placement and shape change of the bladder, leading to effective information about the studied object. This article also shows the developed device, its functions and working principle. The development of such a device and accompanying information technology came about in response to particularly strong market demand for modern technical solutions for urinary tract rehabilitation.

An Ultrasound Tomography Method for Monitoring CO2 Capture Process Involving Stirring and CaCO3 Precipitation.

In this work, an ultrasound computed tomography (USCT) system was employed to investigate the fast-kinetic reactive crystallization process of calcium carbonate. USCT measurements and reconstruction provided key insights into the bulk particle distribution inside the stirred tank reactor and could be used to estimate the settling rate and settling time of the particles. To establish the utility of the USCT system for dynamical crystallization processes, first, the experimental imaging tasks were carried out with the stirred solid beads, as well as the feeding and stirring of the CaCO3 crystals. The feeding region, the mixing process, and the particles settling time could be detected from USCT data. Reactive crystallization experiments for CO2 capture were then conducted. Moreover, there was further potential for quantitative characterization of the suspension density in this process. USCT-based reconstructions were investigated for several experimental scenarios and operating conditions. This study demonstrates a real-time monitoring and fault detection application of USCT for reactive crystallization processes. As a robust noninvasive and nonintrusive tool, real-time signal analysis and reconstruction can be beneficial in the development of monitoring and control systems with real-world applications for crystallization processes. A diverse range of experimental studies shown here demonstrate the versatility of the USCT system in process application, hoping to unlock the commercial and industrial utility of the USCT devices.

Electrical Tomography Reconstruction Using Reconfigurable Waveforms in a FPGA.

The principal objective of this research is to conceive a mobile system based on electrical tomography for subsurface imaging and monitoring in order to enable simultaneous recording of electrical potentials of cardiac and pulmonary activity. For an exploration of excitation waveforms in electrical tomography, specialized hardware is required. As the main principle of tomography is the measurement of electrical perturbations on an unknown object, it is crucial to synchronize excitation and sensing processes in a very precise way for the purpose of acquiring meaningful data. To cope with this problem, an FPGA device is used, with an architecture that allows us to trigger excitation signals and to read sensed data simultaneously via independent processes that share the same clock. In this way, waveform reconfiguration on frequency and shape can be provided and studied. The system is connected to a standard microcontroller SoC with a simple API that allows for IoT capabilities for on-line operation and tracking, given that the design is targeted for in vivo medical monitoring. As a result of the research work, a measuring device was developed, the surface data analyzed and the image was reconstructed using the selected configuration.

Ultrasonic Time-of-Flight Computed Tomography for Investigation of Batch Crystallisation Processes.

Crystallisation is a crucial step in many industrial processes. Many sensors are being investigated for monitoring such processes to enhance the efficiency of them. Ultrasound techniques have been used for particle sizing characterization of liquid suspensions, in crystallisation process. An ultrasound tomography system with an array of ultrasound sensors can provide spatial information inside the process when compared to single-measurement systems. In this study, the batch crystallisation experiments have been conducted in a lab-scale reactor in calcium carbonate crystallisation. Real-time ultrasound tomographic imaging is done via a contactless ultrasound tomography sensor array. The effect of the injection rate and the stirring speed was considered as two control parameters in these crystallisation functions. Transmission mode ultrasound tomography comprises 32 piezoelectric transducers with central frequency of 40 kHz has been used. The process-based experimental investigation shows the capability of the proposed ultrasound tomography system for crystallisation process monitoring. Information on process dynamics, as well as process malfunction, can be obtained via the ultrasound tomography system.

Quality Assessment of the Neural Algorithms on the Example of EIT-UST Hybrid Tomography.

The paper presents the results of research on the hybrid industrial tomograph electrical impedance tomography (EIT) and ultrasonic tomography (UST) (EIT-UST), operating on the basis of electrical and ultrasonic data. The emphasis of the research was placed on the algorithmic domain. However, it should be emphasized that all hardware components of the hybrid tomograph, including electronics, sensors and transducers, have been designed and mostly made in the Netrix S.A. laboratory. The test object was a tank filled with water with several dozen percent concentration. As part of the study, the original multiple neural networks system was trained, the characteristic feature of which is the generation of each of the individual pixels of the tomographic image, using an independent artificial neural network (ANN), with the input vector for all ANNs being the same. Despite the same measurement vector, each of the ANNs generates its own independent output value for a given tomogram pixel, because, during training, the networks get their respective weights and biases. During the tests, the results of three tomographic methods were compared: EIT, UST and EIT-UST hybrid. The results confirm that the use of heterogeneous tomographic systems (hybrids) increases the reliability of reconstruction in various measuring cases, which is used to solve quality problems in managing production processes.

A Quantitative Ultrasonic Travel-Time Tomography to Investigate Liquid Elaborations in Industrial Processes.

This work presents an ultrasound tomography imaging system and method for quantitative mapping of the sound speed in liquid masses. It is highly desirable to be able to inspect vessel fluid mass distribution, notably in the chemical and food industrial operations. Optimization of industrial reactors has been crucial to the improvement of industrial processes. There is a great need to investigate how and if tomographic imaging sensors could aid the automatic control of these process tanks. Single-measurement ultrasound techniques and especially spectrometric methods have been a subject of study of industrial applications. Tomographic systems provide key multi-dimensional and spatial information when compared to the well-established single-channel measurement system. Recently, ultrasound tomography has attracted a great deal of interest in a wide spectrum of industrial applications. The system has been designed as 32 piezoelectric ring-array positioned in a 30 cm tank, with an excitation frequency of 40 kHz. Two-dimensional transmission travel-time tomography was developed to reconstruct the fluid mass distributions. Prior experiments are mainly based on inclusions of a few centimetres and on a liquid solution of different concentrations. They have been conducted to test the spatial and quantitative resolution of the ultrasound imaging device. Analysing the reconstructed images, it is possible to provide accurate spatial resolution with low position errors. The system also demonstrated inclusion movement with a temporal resolution of 4 frames per second (fps) in dynamical imaging sense. Sound velocity quantitative imaging was developed for the investigation of ultrasonic propagation in different liquids. This work, for the first time, shows how quantitative sound velocity imaging using transmission mode time of flight data could be used to characterize liquid density distribution of industrial reactors. The results suggest that ultrasound tomography can be used to quantitatively monitor important process parameters.

Logistic Regression for Machine Learning in Process Tomography.

The main goal of the research presented in this paper was to develop a refined machine learning algorithm for industrial tomography applications. The article presents algorithms based on logistic regression in relation to image reconstruction using electrical impedance tomography (EIT) and ultrasound transmission tomography (UST). The test object was a tank filled with water in which reconstructed objects were placed. For both EIT and UST, a novel approach was used in which each pixel of the output image was reconstructed by a separately trained prediction system. Therefore, it was necessary to use many predictive systems whose number corresponds to the number of pixels of the output image. Thanks to this approach the under-completed problem was changed to an over-completed one. To reduce the number of predictors in logistic regression by removing irrelevant and mutually correlated entries, the elastic net method was used. The developed algorithm that reconstructs images pixel-by-pixel is insensitive to the shape, number and position of the reconstructed objects. In order to assess the quality of mappings obtained thanks to the new algorithm, appropriate metrics were used: compatibility ratio (CR) and relative error (RE). The obtained results enabled the assessment of the usefulness of logistic regression in the reconstruction of EIT and UST images.

Comparison of Selected Machine Learning Algorithms for Industrial Electrical Tomography.

The main goal of this work was to compare the selected machine learning methods with the classic deterministic method in the industrial field of electrical impedance tomography. The research focused on the development and comparison of algorithms and models for the analysis and reconstruction of data using electrical tomography. The novelty was the use of original machine learning algorithms. Their characteristic feature is the use of many separately trained subsystems, each of which generates a single pixel of the output image. Artificial Neural Network (ANN), LARS and Elastic net methods were used to solve the inverse problem. These algorithms have been modified by a corresponding increase in equations (multiply) for electrical impedance tomography using the finite element method grid. The Gauss-Newton method was used as a reference to machine learning methods. The algorithms were trained using learning data obtained through computer simulation based on real models. The results of the experiments showed that in the considered cases the best quality of reconstructions was achieved by ANN. At the same time, ANN was the slowest in terms of both the training process and the speed of image generation. Other machine learning methods were comparable with the deterministic Gauss-Newton method and with each other.

A Non-Destructive System Based on Electrical Tomography and Machine Learning to Analyze the Moisture of Buildings.

This article presents the results of research on a new method of spatial analysis of walls and buildings moisture. Due to the fact that destructive methods are not suitable for historical buildings of great architectural significance, a non-destructive method based on electrical tomography has been adopted. A hybrid tomograph with special sensors was developed for the measurements. This device enables the acquisition of data, which are then reconstructed by appropriately developed methods enabling spatial analysis of wet buildings. Special electrodes that ensure good contact with the surface of porous building materials such as bricks and cement were introduced. During the research, a group of algorithms enabling supervised machine learning was analyzed. They have been used in the process of converting input electrical values into conductance depicted by the output image pixels. The conductance values of individual pixels of the output vector made it possible to obtain images of the interior of building walls as both flat intersections (2D) and spatial (3D) images. The presented group of algorithms has a high application value. The main advantages of the new methods are: high accuracy of imaging, low costs, high processing speed, ease of application to walls of various thickness and irregular surface. By comparing the results of tomographic reconstructions, the most efficient algorithms were identified.