Cargando…

Automatic Identification of Ultrasound Images of the Tibial Nerve in Different Ankle Positions Using Deep Learning

Peripheral nerve tension is known to be related to the pathophysiology of neuropathy; however, assessing this tension is difficult in a clinical setting. In this study, we aimed to develop a deep learning algorithm for the automatic assessment of tibial nerve tension using B-mode ultrasound imaging....

Descripción completa

Detalles Bibliográficos
Autores principales: Kawanishi, Kengo, Kakimoto, Akihiro, Anegawa, Keisuke, Tsutsumi, Masahiro, Yamaguchi, Isao, Kudo, Shintarou
Formato: Online Artículo Texto
Lenguaje:English
Publicado: MDPI 2023
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10222534/
https://www.ncbi.nlm.nih.gov/pubmed/37430769
http://dx.doi.org/10.3390/s23104855
_version_ 1785049721812811776
author Kawanishi, Kengo
Kakimoto, Akihiro
Anegawa, Keisuke
Tsutsumi, Masahiro
Yamaguchi, Isao
Kudo, Shintarou
author_facet Kawanishi, Kengo
Kakimoto, Akihiro
Anegawa, Keisuke
Tsutsumi, Masahiro
Yamaguchi, Isao
Kudo, Shintarou
author_sort Kawanishi, Kengo
collection PubMed
description Peripheral nerve tension is known to be related to the pathophysiology of neuropathy; however, assessing this tension is difficult in a clinical setting. In this study, we aimed to develop a deep learning algorithm for the automatic assessment of tibial nerve tension using B-mode ultrasound imaging. To develop the algorithm, we used 204 ultrasound images of the tibial nerve in three positions: the maximum dorsiflexion position and −10° and −20° plantar flexion from maximum dorsiflexion. The images were taken of 68 healthy volunteers who did not have any abnormalities in the lower limbs at the time of testing. The tibial nerve was manually segmented in all images, and 163 cases were automatically extracted as the training dataset using U-Net. Additionally, convolutional neural network (CNN)-based classification was performed to determine each ankle position. The automatic classification was validated using five-fold cross-validation from the testing data composed of 41 data points. The highest mean accuracy (0.92) was achieved using manual segmentation. The mean accuracy of the full auto-classification of the tibial nerve at each ankle position was more than 0.77 using five-fold cross-validation. Thus, the tension of the tibial nerve can be accurately assessed with different dorsiflexion angles using an ultrasound imaging analysis with U-Net and a CNN.
format Online
Article
Text
id pubmed-10222534
institution National Center for Biotechnology Information
language English
publishDate 2023
publisher MDPI
record_format MEDLINE/PubMed
spelling pubmed-102225342023-05-28 Automatic Identification of Ultrasound Images of the Tibial Nerve in Different Ankle Positions Using Deep Learning Kawanishi, Kengo Kakimoto, Akihiro Anegawa, Keisuke Tsutsumi, Masahiro Yamaguchi, Isao Kudo, Shintarou Sensors (Basel) Communication Peripheral nerve tension is known to be related to the pathophysiology of neuropathy; however, assessing this tension is difficult in a clinical setting. In this study, we aimed to develop a deep learning algorithm for the automatic assessment of tibial nerve tension using B-mode ultrasound imaging. To develop the algorithm, we used 204 ultrasound images of the tibial nerve in three positions: the maximum dorsiflexion position and −10° and −20° plantar flexion from maximum dorsiflexion. The images were taken of 68 healthy volunteers who did not have any abnormalities in the lower limbs at the time of testing. The tibial nerve was manually segmented in all images, and 163 cases were automatically extracted as the training dataset using U-Net. Additionally, convolutional neural network (CNN)-based classification was performed to determine each ankle position. The automatic classification was validated using five-fold cross-validation from the testing data composed of 41 data points. The highest mean accuracy (0.92) was achieved using manual segmentation. The mean accuracy of the full auto-classification of the tibial nerve at each ankle position was more than 0.77 using five-fold cross-validation. Thus, the tension of the tibial nerve can be accurately assessed with different dorsiflexion angles using an ultrasound imaging analysis with U-Net and a CNN. MDPI 2023-05-18 /pmc/articles/PMC10222534/ /pubmed/37430769 http://dx.doi.org/10.3390/s23104855 Text en © 2023 by the authors. https://creativecommons.org/licenses/by/4.0/Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
spellingShingle Communication
Kawanishi, Kengo
Kakimoto, Akihiro
Anegawa, Keisuke
Tsutsumi, Masahiro
Yamaguchi, Isao
Kudo, Shintarou
Automatic Identification of Ultrasound Images of the Tibial Nerve in Different Ankle Positions Using Deep Learning
title Automatic Identification of Ultrasound Images of the Tibial Nerve in Different Ankle Positions Using Deep Learning
title_full Automatic Identification of Ultrasound Images of the Tibial Nerve in Different Ankle Positions Using Deep Learning
title_fullStr Automatic Identification of Ultrasound Images of the Tibial Nerve in Different Ankle Positions Using Deep Learning
title_full_unstemmed Automatic Identification of Ultrasound Images of the Tibial Nerve in Different Ankle Positions Using Deep Learning
title_short Automatic Identification of Ultrasound Images of the Tibial Nerve in Different Ankle Positions Using Deep Learning
title_sort automatic identification of ultrasound images of the tibial nerve in different ankle positions using deep learning
topic Communication
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10222534/
https://www.ncbi.nlm.nih.gov/pubmed/37430769
http://dx.doi.org/10.3390/s23104855
work_keys_str_mv AT kawanishikengo automaticidentificationofultrasoundimagesofthetibialnerveindifferentanklepositionsusingdeeplearning
AT kakimotoakihiro automaticidentificationofultrasoundimagesofthetibialnerveindifferentanklepositionsusingdeeplearning
AT anegawakeisuke automaticidentificationofultrasoundimagesofthetibialnerveindifferentanklepositionsusingdeeplearning
AT tsutsumimasahiro automaticidentificationofultrasoundimagesofthetibialnerveindifferentanklepositionsusingdeeplearning
AT yamaguchiisao automaticidentificationofultrasoundimagesofthetibialnerveindifferentanklepositionsusingdeeplearning
AT kudoshintarou automaticidentificationofultrasoundimagesofthetibialnerveindifferentanklepositionsusingdeeplearning