GPU-accelerated Phased-Array Ultrasound Scanner for Integration of AI/ML methods
e-Journal of Nondestructive Testing · 2026-07 · 31 (7)
DOI: 10.58286/33412 ↗At a glance
Journal article in e-Journal of Nondestructive Testing, 2026-07. Written by Marcin Lewandowski and 3 co-authors.
Abstract (original language)
The integration of artificial intelligence and machine learning (AI/ML) into ultrasonic non-destructive testing (NDT) is a key requirement for meeting the demands of Industry 4.0, including automation, robotization, and data-driven inspection. Conventional Phased-Array ultrasonic testing (PAUT) systems often rely on operator-dependent interpretation and are limited by insufficient processing power for real-time execution of advanced imaging and AI/ML algorithms. This work presents a GPU-accelerated Phased-Array ultrasound scanner designed as an open platform for real-time full matrix capture (FMC), total focusing method (TFM) imaging, and on-device AI/ML inference. The us4NDT system combines a fully programmable ultrasonic front-end with an embedded NVIDIA Jetson AGX Orin GPU module, enabling high-throughput data acquisition and low-latency processing. The platform supports FMC data rates up to 3 GB/s per module and executes GPU-based TFM reconstruction alongside custom processing kernels and neural network inference. To validate the system, convolutional neural network models were implemented for defect detection and sizing, including custom YOLO model for object detection. Preliminary results demonstrate real-time TFM imaging combined with AI/ML-based defect detection within a compact, industrial-grade device. The proposed system provides a flexible and high-performance solution for intelligent ultrasonic inspection and serves as a foundation for autonomous and AI-enhanced NDT applications.
Cite
APA
Lewandowski, M., Walczak, M., Jarosik, P., & Karwat, P. (2026). GPU-accelerated Phased-Array Ultrasound Scanner for Integration of AI/ML methods. e-Journal of Nondestructive Testing, 31(7). https://doi.org/10.58286/33412
Vancouver
Lewandowski M, Walczak M, Jarosik P, Karwat P. GPU-accelerated Phased-Array Ultrasound Scanner for Integration of AI/ML methods. e-Journal of Nondestructive Testing. 2026;31(7). doi:10.58286/33412
BibTeX
@article{Lewandowski2026,
author = {Lewandowski, Marcin and Walczak, Mateusz and Jarosik, Piotr and Karwat, Piotr},
title = {GPU-accelerated Phased-Array Ultrasound Scanner for Integration of AI/ML methods},
journal = {e-Journal of Nondestructive Testing},
year = {2026},
volume = {31},
number = {7},
doi = {10.58286/33412}
}
EUVIMED