Register or Log In

International Society for Industrial Process Tomography

11th World Congress on Industrial Process Tomography

AI-powered object detection in triple-channel ultrasound tomography


B. Baran1*, P. Szyszka2, M. Gołąbek1, D. Majerek2, D. Wójcik1,3, T. Rymarczyk1,3

1Research and Development Center, Netrix S.A., Lublin, Poland

2Faculty of Fundamentals of Technology, Lublin University of Technology, Lublin, Poland

3Faculty of Transport and Computer Science, WSEI University, Lublin, Poland

*Email: bartlomiej.baran@netrix.com.pl


ABSTRACT

Ultrasound tomography is one of the leading imaging methods, it can be found in various sectors, e.g. in medicine, industry, or it can be used as a method for nondestructive testing. In this article, we explore the possibility of object detection by ultrasound waves with the support of artificial intelligence algorithms. The purpose of the research is to predict the position of objects based on the analysis of the analog signal collected by ultrasound transducers. We use recorded analog signals from three channels placed side by side. This approach provides reliable object detection using a small amount of information, contrary to other systems that often use dozens of measurement probes. To demonstrate the robustness of our approach, we have adopted and compared many machine learning algorithms. Our investigation reveals that most of the used models feature very high values of Accuracy, R2, or F1 score functions.


Keywords: Ultrasound Tomography, Machine Learning, Artificial Intelligence, Localization, Reconstruction

Industrial Application: General

Sign-in to access the full text