Random Fourier Features-Based Deep Learning Improvement with Class Activation Interpretability for Nerve Structure Segmentation
Peripheral nerve blocking (PNB) is a standard procedure to support regional anesthesia. Still, correct localization of the nerve’s structure is needed to avoid adverse effects; thereby, ultrasound images are used as an aid approach. In addition, image-based automatic nerve segmentation from deep lea...
Guardado en:
Autores principales: | , , , , |
---|---|
Formato: | article |
Lenguaje: | EN |
Publicado: |
MDPI AG
2021
|
Materias: | |
Acceso en línea: | https://doaj.org/article/6a121e7284ca48b89d1884eee2c25d78 |
Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
id |
oai:doaj.org-article:6a121e7284ca48b89d1884eee2c25d78 |
---|---|
record_format |
dspace |
spelling |
oai:doaj.org-article:6a121e7284ca48b89d1884eee2c25d782021-11-25T18:58:56ZRandom Fourier Features-Based Deep Learning Improvement with Class Activation Interpretability for Nerve Structure Segmentation10.3390/s212277411424-8220https://doaj.org/article/6a121e7284ca48b89d1884eee2c25d782021-11-01T00:00:00Zhttps://www.mdpi.com/1424-8220/21/22/7741https://doaj.org/toc/1424-8220Peripheral nerve blocking (PNB) is a standard procedure to support regional anesthesia. Still, correct localization of the nerve’s structure is needed to avoid adverse effects; thereby, ultrasound images are used as an aid approach. In addition, image-based automatic nerve segmentation from deep learning methods has been proposed to mitigate attenuation and speckle noise ultrasonography issues. Notwithstanding, complex architectures highlight the region of interest lacking suitable data interpretability concerning the learned features from raw instances. Here, a kernel-based deep learning enhancement is introduced for nerve structure segmentation. In a nutshell, a random Fourier features-based approach was utilized to complement three well-known semantic segmentation architectures, e.g., fully convolutional network, U-net, and ResUnet. Moreover, two ultrasound image datasets for PNB were tested. Obtained results show that our kernel-based approach provides a better generalization capability from image segmentation-based assessments on different nerve structures. Further, for data interpretability, a semantic segmentation extension of the GradCam++ for class-activation mapping was used to reveal relevant learned features separating between nerve and background. Thus, our proposal favors both straightforward (shallow) and complex architectures (deeper neural networks).Cristian Alfonso Jimenez-CastañoAndrés Marino Álvarez-MezaOscar David Aguirre-OspinaDavid Augusto Cárdenas-PeñaÁlvaro Angel Orozco-GutiérrezMDPI AGarticlenerve structure segmentationultrasound imagesdeep learningrandom Fourier featuresclass activation mappingChemical technologyTP1-1185ENSensors, Vol 21, Iss 7741, p 7741 (2021) |
institution |
DOAJ |
collection |
DOAJ |
language |
EN |
topic |
nerve structure segmentation ultrasound images deep learning random Fourier features class activation mapping Chemical technology TP1-1185 |
spellingShingle |
nerve structure segmentation ultrasound images deep learning random Fourier features class activation mapping Chemical technology TP1-1185 Cristian Alfonso Jimenez-Castaño Andrés Marino Álvarez-Meza Oscar David Aguirre-Ospina David Augusto Cárdenas-Peña Álvaro Angel Orozco-Gutiérrez Random Fourier Features-Based Deep Learning Improvement with Class Activation Interpretability for Nerve Structure Segmentation |
description |
Peripheral nerve blocking (PNB) is a standard procedure to support regional anesthesia. Still, correct localization of the nerve’s structure is needed to avoid adverse effects; thereby, ultrasound images are used as an aid approach. In addition, image-based automatic nerve segmentation from deep learning methods has been proposed to mitigate attenuation and speckle noise ultrasonography issues. Notwithstanding, complex architectures highlight the region of interest lacking suitable data interpretability concerning the learned features from raw instances. Here, a kernel-based deep learning enhancement is introduced for nerve structure segmentation. In a nutshell, a random Fourier features-based approach was utilized to complement three well-known semantic segmentation architectures, e.g., fully convolutional network, U-net, and ResUnet. Moreover, two ultrasound image datasets for PNB were tested. Obtained results show that our kernel-based approach provides a better generalization capability from image segmentation-based assessments on different nerve structures. Further, for data interpretability, a semantic segmentation extension of the GradCam++ for class-activation mapping was used to reveal relevant learned features separating between nerve and background. Thus, our proposal favors both straightforward (shallow) and complex architectures (deeper neural networks). |
format |
article |
author |
Cristian Alfonso Jimenez-Castaño Andrés Marino Álvarez-Meza Oscar David Aguirre-Ospina David Augusto Cárdenas-Peña Álvaro Angel Orozco-Gutiérrez |
author_facet |
Cristian Alfonso Jimenez-Castaño Andrés Marino Álvarez-Meza Oscar David Aguirre-Ospina David Augusto Cárdenas-Peña Álvaro Angel Orozco-Gutiérrez |
author_sort |
Cristian Alfonso Jimenez-Castaño |
title |
Random Fourier Features-Based Deep Learning Improvement with Class Activation Interpretability for Nerve Structure Segmentation |
title_short |
Random Fourier Features-Based Deep Learning Improvement with Class Activation Interpretability for Nerve Structure Segmentation |
title_full |
Random Fourier Features-Based Deep Learning Improvement with Class Activation Interpretability for Nerve Structure Segmentation |
title_fullStr |
Random Fourier Features-Based Deep Learning Improvement with Class Activation Interpretability for Nerve Structure Segmentation |
title_full_unstemmed |
Random Fourier Features-Based Deep Learning Improvement with Class Activation Interpretability for Nerve Structure Segmentation |
title_sort |
random fourier features-based deep learning improvement with class activation interpretability for nerve structure segmentation |
publisher |
MDPI AG |
publishDate |
2021 |
url |
https://doaj.org/article/6a121e7284ca48b89d1884eee2c25d78 |
work_keys_str_mv |
AT cristianalfonsojimenezcastano randomfourierfeaturesbaseddeeplearningimprovementwithclassactivationinterpretabilityfornervestructuresegmentation AT andresmarinoalvarezmeza randomfourierfeaturesbaseddeeplearningimprovementwithclassactivationinterpretabilityfornervestructuresegmentation AT oscardavidaguirreospina randomfourierfeaturesbaseddeeplearningimprovementwithclassactivationinterpretabilityfornervestructuresegmentation AT davidaugustocardenaspena randomfourierfeaturesbaseddeeplearningimprovementwithclassactivationinterpretabilityfornervestructuresegmentation AT alvaroangelorozcogutierrez randomfourierfeaturesbaseddeeplearningimprovementwithclassactivationinterpretabilityfornervestructuresegmentation |
_version_ |
1718410466338799616 |