vmaitComputational Mathematics and Information TechnologiesComputational Mathematics and Information Technologies2587-8999Донской государственный технический университет10.23947/2587-8999-2024-8-2-60-67vmait-159Research ArticleInformation Technologies (Информационные технологии)Определение границы раздела сред по трёхмерным данным матричного ультразвукового датчика с использованием свёрточных нейронных сетейLocating the Interface between Different Media Based on Matrix Ultrasonic Sensor Data Using Convolutional Neural Networkshttps://orcid.org/0000-0003-0513-1395ВасюковА. В.VasyukovA. V.

Васюков Алексей Викторович, старший научный сотрудник кафедры информатики и вычислительной математики; кандидат физико-математических наук

141701, Долгопрудный, Институтский переулок, 9

Alexey V. Vasyukov, Senior Research Fellow at the Department of Informatics and Compu-tational Mathematics

9, Institutsky Lane, Dolgoprudny, 141701

vasyukov.av@mipt.ruМосковский физико-технический институт (национальный исследовательский университет)РоссияМоscow Institute of Physics and Technology (National Research University)Russian Federation202405072024826067Copyright © Васюков А.В., 20242024Васюков А.В.Vasyukov A.V.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://www.cmit-journal.ru/jour/article/view/159

Введение. Работа посвящена моделированию процесса ультразвукового медицинского исследования в гетерогенной среде, в которой присутствуют области с существенно разной скоростью звука. Такие постановки задач возникают, например, при визуализации структур мозга через череп. Целью данной работы является сравнение возможных подходов к определению границы раздела акустически контрастных сред с использованием свёрточных нейронных сетей.Материалы и методы. В работе выполняется численное моделирование прямой задачи — получение синтетических расчётных ультразвуковых изображений по известной геометрии и реологии области, а также параметрам датчика. На расчётных изображениях воспроизводятся искажения и артефакты, типичные для постановок со стенкой черепа. Для решения обратной задачи определения границы раздела сред по сигналу с датчика используются свёрточные нейронные сети 2D и 3D структуры, следующие общей архитектуре UNet. Сети обучаются на наборах расчётных данных, после чего тестируются на отдельных примерах, не использованных при обучении.Результаты исследования. Получены расчётные B-сканы для характерных постановок. Показана возможность локализации границы аберратора с хорошим качеством как для 2D, так и для 3D свёрточных сетей. Показано более высокое качество результата для 3D сетей в случае наличия значительного шума и артефактов во входных данных. Установлено, что сеть 3D архитектуры может обеспечить получение формы границы раздела сред за 0,1 секунды.Обсуждение и заключения. Результаты работы могут быть использованы для развития технологий транскраниального ультразвукового исследования. Быстрая локализация границы стенки черепа может быть включена в алгоритмы построения изображения для компенсации искажений, вызванных различием скоростей звука в костных и в мягких тканях.

Introduction. The study focuses on modelling the process of ultrasound medical examination in a heterogeneous environment with regions of significantly different sound speeds. Such scenarios typically arise when visualizing brain structures through the skull. The aim of this work is to compare possible approaches to determining the interface between acoustically contrasting media using convolutional neural networks.Materials and Methods. Numerical modelling of the direct problem is performed, obtaining synthetic calculated ultrasonic images based on known geometry and rheology of the area as well as sensor parameters. The calculated images reproduce distortions and artifacts typical for setups involving the skull wall. Convolutional neural networks of 2D and 3D structures following the UNet architecture are used to solve the inverse problem of determining the interface between media based on a sensor signal. The networks are trained on computational datasets and then tested on individual samples not used in training.Results. Numerical B-scans for characteristic setups were obtained. The possibility of localizing the aberrator boundary with good quality for both 2D and 3D convolutional networks was demonstrated. A higher quality result was obtained for the 3D network in the presence of significant noise and artifacts in the input data. It was established that the 3D architecture network can provide the shape of the interface between media in 0.1 seconds.Discussion and Conclusions. The results can be used for the development of transcranial ultrasound technologies. Rapid localization of the skull boundary can be incorporated into imaging algorithms to compensate for distortions caused by differences in sound velocities in bone and soft tissues.

транскраниальное УЗИматричный датчикаберрацииматематическое моделированиесеточно-характеристический методсверточные сетиtranscranial ultrasoundmatrix probeaberrationsmathematical modellinggrid-characteristic methodconvolutional networksРабота выполнена при финансовой поддержке Российского научного фонда (код проекта 22-11-00142).The work was carried out with financial support from the Russian Science Foundation (project 22-11-00142).ReferencesBeklemysheva K.A., Grigoriev G.K., Kulberg N.S., Petrov I.B., Vasyukov A.V., Vassilevski Y.V. Numerical simulation of aberrated medical ultrasound signals. Russian Journal of Numerical Analysis and Mathematical Modelling. 2018;33(5):277–288. https://doi.org/10.1515/rnam-2018-0023Beklemysheva K.A., Grigoriev G.K., Kulberg N.S., Petrov I.B., Vasyukov A.V., Vassilevski Y.V. Numerical simulation of aberrated medical ultrasound signals. Russian Journal of Numerical Analysis and Mathematical Modelling. 2018;33(5):277–288. https://doi.org/10.1515/rnam-2018-0023Perdios D., Vonlanthen M., Martinez F., Arditi M., Thiran J.P. Single-shot CNN-based ultrasound imaging with sparse linear arrays. In: 2020 IEEE International Ultrasonics Symposium (IUS). Las Vegas, NV, USA; 2020. P. 1‒4. https://doi.org/10.1109/IUS46767.2020.9251442Perdios D., Vonlanthen M., Martinez F., Arditi M., Thiran J.P. Single-shot CNN-based ultrasound imaging with sparse linear arrays. In: 2020 IEEE International Ultrasonics Symposium (IUS). Las Vegas, NV, USA; 2020. P. 1‒4. https://doi.org/10.1109/IUS46767.2020.9251442Patel D., Tibrewala R., Vega A., Dong L., Hugenberg N., Oberai A. Circumventing the solution of inverse problems in mechanics through deep learning: Application to elasticity imaging. Computer Methods in Applied Mechanics and Engineering. 2019;353:448–466. https://doi.org/10.1016/j.cma.2019.04.045Patel D., Tibrewala R., Vega A., Dong L., Hugenberg N., Oberai A. Circumventing the solution of inverse problems in mechanics through deep learning: Application to elasticity imaging. Computer Methods in Applied Mechanics and Engineering. 2019;353:448–466. https://doi.org/10.1016/j.cma.2019.04.045Hongya Lu, Haifeng Wang, Qianqian Zhang, Sang Won Yoon, Daehan Won. A 3D Convolutional Neural Network for Volumetric Image Semantic Segmentation. Procedia Manufacturing. 2019;39:422–428. https://doi.org/10.1016/j.promfg.2020.01.386Hongya Lu, Haifeng Wang, Qianqian Zhang, Sang Won Yoon, Daehan Won. A 3D Convolutional Neural Network for Volumetric Image Semantic Segmentation. Procedia Manufacturing. 2019;39:422–428. https://doi.org/10.1016/j.promfg.2020.01.386Potočnik B., Šavc M. Deeply-Supervised 3D Convolutional Neural Networks for Automated Ovary and Follicle Detection from Ultrasound Volumes. Applied Sciences. 2022;12(3):1246. https://doi.org/10.3390/app12031246Potočnik B., Šavc M. Deeply-Supervised 3D Convolutional Neural Networks for Automated Ovary and Follicle Detection from Ultrasound Volumes. Applied Sciences. 2022;12(3):1246. https://doi.org/10.3390/app12031246Brown K., Dormer J., Fei B., Hoyt K. Deep 3D convolutional neural networks for fast super-resolution ultrasound imaging. Proceedings SPIE 10955, Medical Imaging 2019: Ultrasonic Imaging and Tomography. 2019;10955:1095502. https://doi.org/10.1117/12.2511897Brown K., Dormer J., Fei B., Hoyt K. Deep 3D convolutional neural networks for fast super-resolution ultrasound imaging. Proceedings SPIE 10955, Medical Imaging 2019: Ultrasonic Imaging and Tomography. 2019;10955:1095502. https://doi.org/10.1117/12.2511897Mast T.D., Hinkelman L.M., Metlay L.A., Orr M.J., Waag R.C. Simulation of ultrasonic pulse propagation, distortion, and attenuation in the human chest wall. J. Acoust. Soc. Amer. 1999;6:3665–3677. https://doi.org/10.1121/1.428209Mast T.D., Hinkelman L.M., Metlay L.A., Orr M.J., Waag R.C. Simulation of ultrasonic pulse propagation, distortion, and attenuation in the human chest wall. J. Acoust. Soc. Amer. 1999;6:3665–3677. https://doi.org/10.1121/1.428209Madsen E.L., Sathoff H.J., Zagzebski J.A. Ultrasonic shear wave properties of soft tissues and tissuelike materials. J. Acoust. Soc. Am. 1983;74(5):1346–1355. https://doi.org/10.1121/1.390158Madsen E.L., Sathoff H.J., Zagzebski J.A. Ultrasonic shear wave properties of soft tissues and tissuelike materials. J. Acoust. Soc. Am. 1983;74(5):1346–1355. https://doi.org/10.1121/1.390158Vassilevski Y.V., Beklemysheva K.A., Grigoriev G.K., Kulberg N.S., Petrov I.B., Vasyukov A.V. Numerical modelling of medical ultrasound: phantom-based verification. Russian Journal of Numerical Analysis and Mathematical Modelling. 2017;32(5):339–346. https://doi.org/10.1515/rnam-2017-0032Vassilevski Y.V., Beklemysheva K.A., Grigoriev G.K., Kulberg N.S., Petrov I.B., Vasyukov A.V. Numerical modelling of medical ultrasound: phantom-based verification. Russian Journal of Numerical Analysis and Mathematical Modelling. 2017;32(5):339–346. https://doi.org/10.1515/rnam-2017-0032Ronneberger O., Fischer P., Brox T. U-Net: Convolutional Networks for Biomedical Image Segmentation. Lecture Notes in Computer Science. 2015;9351:234–241. https://doi.org/10.1007/978-3-319-24574-4_28Ronneberger O., Fischer P., Brox T. U-Net: Convolutional Networks for Biomedical Image Segmentation. Lecture Notes in Computer Science. 2015;9351:234–241. https://doi.org/10.1007/978-3-319-24574-4_28Paserin O., Mulpuri K., Cooper A., Abugharbieh R., Hodgson A. Improving 3D Ultrasound Scan Adequacy Classification Using a Three-Slice Convolutional Neural Network Architecture. In: CAOS 2018 (EPiC Series in Health Sciences Vol 2). Beijing, China; 2018. P. 152–156. https://doi.org/10.29007/2tctPaserin O., Mulpuri K., Cooper A., Abugharbieh R., Hodgson A. Improving 3D Ultrasound Scan Adequacy Classification Using a Three-Slice Convolutional Neural Network Architecture. In: CAOS 2018 (EPiC Series in Health Sciences Vol 2). Beijing, China; 2018. P. 152–156. https://doi.org/10.29007/2tctJiang M., Spence J.D., Chiu B. Segmentation of 3D ultrasound carotid vessel wall using U-Net and segmentation average network. In: 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC). Montreal, QC, Canada; 2020. P. 2043–2046. https://doi.org/10.1109/EMBC44109.2020.9175975Jiang M., Spence J.D., Chiu B. Segmentation of 3D ultrasound carotid vessel wall using U-Net and segmentation average network. In: 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC). Montreal, QC, Canada; 2020. P. 2043–2046. https://doi.org/10.1109/EMBC44109.2020.9175975Zheng Y., Liu D., Georgescu B., Nguyen H., Comaniciu D. 3D deep learning for efficient and robust landmark detection in volumetric data. Lecture Notes in Computer Science. 2015;9349:565–572. https://doi.org/10.1007/978-3-319-24553-9_69Zheng Y., Liu D., Georgescu B., Nguyen H., Comaniciu D. 3D deep learning for efficient and robust landmark detection in volumetric data. Lecture Notes in Computer Science. 2015;9349:565–572. https://doi.org/10.1007/978-3-319-24553-9_69Ghimire K., Chen Q., Feng X. Patch-Based 3D UNet for Head and Neck Tumor Segmentation with an Ensemble of Conventional and Dilated Convolutions. Lecture Notes in Computer Science. 2021;12603:78–84. https://doi.org/10.1007/978-3-030-67194-5_9Ghimire K., Chen Q., Feng X. Patch-Based 3D UNet for Head and Neck Tumor Segmentation with an Ensemble of Conventional and Dilated Convolutions. Lecture Notes in Computer Science. 2021;12603:78–84. https://doi.org/10.1007/978-3-030-67194-5_9Coupeau P., Fasquel J.B., Mazerand E., Menei P., Montero-Menei C.N., Dinomais M. Patch-based 3D U-Net and transfer learning for longitudinal piglet brain segmentation on MRI. Computer Methods and Programs in Biomedicine. 2022;214:106563. https://doi.org/10.1016/j.cmpb.2021.106563Coupeau P., Fasquel J.B., Mazerand E., Menei P., Montero-Menei C.N., Dinomais M. Patch-based 3D U-Net and transfer learning for longitudinal piglet brain segmentation on MRI. Computer Methods and Programs in Biomedicine. 2022;214:106563. https://doi.org/10.1016/j.cmpb.2021.106563

The authors declare that there are no conflicts of interest present.