Ten Fast Transfer Learning Models for Carotid Ultrasound Plaque Tissue Characterization in Augmentation Framework Embedded with Heatmaps for Stroke Risk Stratification

<i>Background and Purpose:</i> Only 1–2% of the internal carotid artery asymptomatic plaques are unstable as a result of >80% stenosis. Thus, unnecessary efforts can be saved if these plaques can be characterized and classified into symptomatic and asymptomatic using non-invasive B-mo...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Skandha S. Sanagala, Andrew Nicolaides, Suneet K. Gupta, Vijaya K. Koppula, Luca Saba, Sushant Agarwal, Amer M. Johri, Manudeep S. Kalra, Jasjit S. Suri
Formato:	article
Lenguaje:	EN
Publicado:	MDPI AG 2021
Materias:	stroke carotid plaque characterization symptomatic vs. asymptomatic artificial intelligence transfer learning heatmaps Medicine (General) R5-920
Acceso en línea:	https://doaj.org/article/328b577cc70340eaa8d8697fe1e80bbb
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:328b577cc70340eaa8d8697fe1e80bbb
record_format	dspace
spelling	oai:doaj.org-article:328b577cc70340eaa8d8697fe1e80bbb2021-11-25T17:21:42ZTen Fast Transfer Learning Models for Carotid Ultrasound Plaque Tissue Characterization in Augmentation Framework Embedded with Heatmaps for Stroke Risk Stratification10.3390/diagnostics111121092075-4418https://doaj.org/article/328b577cc70340eaa8d8697fe1e80bbb2021-11-01T00:00:00Zhttps://www.mdpi.com/2075-4418/11/11/2109https://doaj.org/toc/2075-4418<i>Background and Purpose:</i> Only 1–2% of the internal carotid artery asymptomatic plaques are unstable as a result of >80% stenosis. Thus, unnecessary efforts can be saved if these plaques can be characterized and classified into symptomatic and asymptomatic using non-invasive B-mode ultrasound. Earlier plaque tissue characterization (PTC) methods were machine learning (ML)-based, which used hand-crafted features that yielded lower accuracy and unreliability. The proposed study shows the role of transfer learning (TL)-based deep learning models for PTC. <i>Methods:</i> As pertained weights were used in the supercomputer framework, we hypothesize that transfer learning (TL) provides improved performance compared with deep learning. We applied 11 kinds of artificial intelligence (AI) models, 10 of them were augmented and optimized using TL approaches—a class of Atheromatic™ 2.0 <sub>TL</sub> (AtheroPoint™, Roseville, CA, USA) that consisted of (<b>i–ii</b>) Visual Geometric Group-16, 19 (VGG16, 19); (<b>iii</b>) Inception V3 (IV3); (<b>iv–v</b>) DenseNet121, 169; (<b>vi</b>) XceptionNet; (<b>vii</b>) ResNet50; (<b>viii</b>) MobileNet; (<b>ix</b>) AlexNet; (<b>x</b>) SqueezeNet; and one DL-based (<b>xi</b>) SuriNet-derived from UNet. We benchmark 11 AI models against our earlier deep convolutional neural network (DCNN) model. <i>Results:</i> The best performing TL was MobileNet, with accuracy and area-under-the-curve (AUC) pairs of 96.10 ± 3% and 0.961 (<i>p</i> < 0.0001), respectively. In DL, DCNN was comparable to SuriNet, with an accuracy of 95.66% and 92.7 ± 5.66%, and an AUC of 0.956 (<i>p</i> < 0.0001) and 0.927 (<i>p</i> < 0.0001), respectively. We validated the performance of the AI architectures with established biomarkers such as greyscale median (GSM), fractal dimension (FD), higher-order spectra (HOS), and visual heatmaps. We benchmarked against previously developed Atheromatic™ 1.0 <sub>ML</sub> and showed an improvement of <b>12.9</b>%. <i>Conclusions:</i> TL is a powerful AI tool for PTC into symptomatic and asymptomatic plaques.Skandha S. SanagalaAndrew NicolaidesSuneet K. GuptaVijaya K. KoppulaLuca SabaSushant AgarwalAmer M. JohriManudeep S. KalraJasjit S. SuriMDPI AGarticlestrokecarotid plaque characterizationsymptomatic vs. asymptomaticartificial intelligencetransfer learningheatmapsMedicine (General)R5-920ENDiagnostics, Vol 11, Iss 2109, p 2109 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	stroke carotid plaque characterization symptomatic vs. asymptomatic artificial intelligence transfer learning heatmaps Medicine (General) R5-920
spellingShingle	stroke carotid plaque characterization symptomatic vs. asymptomatic artificial intelligence transfer learning heatmaps Medicine (General) R5-920 Skandha S. Sanagala Andrew Nicolaides Suneet K. Gupta Vijaya K. Koppula Luca Saba Sushant Agarwal Amer M. Johri Manudeep S. Kalra Jasjit S. Suri Ten Fast Transfer Learning Models for Carotid Ultrasound Plaque Tissue Characterization in Augmentation Framework Embedded with Heatmaps for Stroke Risk Stratification
description	<i>Background and Purpose:</i> Only 1–2% of the internal carotid artery asymptomatic plaques are unstable as a result of >80% stenosis. Thus, unnecessary efforts can be saved if these plaques can be characterized and classified into symptomatic and asymptomatic using non-invasive B-mode ultrasound. Earlier plaque tissue characterization (PTC) methods were machine learning (ML)-based, which used hand-crafted features that yielded lower accuracy and unreliability. The proposed study shows the role of transfer learning (TL)-based deep learning models for PTC. <i>Methods:</i> As pertained weights were used in the supercomputer framework, we hypothesize that transfer learning (TL) provides improved performance compared with deep learning. We applied 11 kinds of artificial intelligence (AI) models, 10 of them were augmented and optimized using TL approaches—a class of Atheromatic™ 2.0 <sub>TL</sub> (AtheroPoint™, Roseville, CA, USA) that consisted of (<b>i–ii</b>) Visual Geometric Group-16, 19 (VGG16, 19); (<b>iii</b>) Inception V3 (IV3); (<b>iv–v</b>) DenseNet121, 169; (<b>vi</b>) XceptionNet; (<b>vii</b>) ResNet50; (<b>viii</b>) MobileNet; (<b>ix</b>) AlexNet; (<b>x</b>) SqueezeNet; and one DL-based (<b>xi</b>) SuriNet-derived from UNet. We benchmark 11 AI models against our earlier deep convolutional neural network (DCNN) model. <i>Results:</i> The best performing TL was MobileNet, with accuracy and area-under-the-curve (AUC) pairs of 96.10 ± 3% and 0.961 (<i>p</i> < 0.0001), respectively. In DL, DCNN was comparable to SuriNet, with an accuracy of 95.66% and 92.7 ± 5.66%, and an AUC of 0.956 (<i>p</i> < 0.0001) and 0.927 (<i>p</i> < 0.0001), respectively. We validated the performance of the AI architectures with established biomarkers such as greyscale median (GSM), fractal dimension (FD), higher-order spectra (HOS), and visual heatmaps. We benchmarked against previously developed Atheromatic™ 1.0 <sub>ML</sub> and showed an improvement of <b>12.9</b>%. <i>Conclusions:</i> TL is a powerful AI tool for PTC into symptomatic and asymptomatic plaques.
format	article
author	Skandha S. Sanagala Andrew Nicolaides Suneet K. Gupta Vijaya K. Koppula Luca Saba Sushant Agarwal Amer M. Johri Manudeep S. Kalra Jasjit S. Suri
author_facet	Skandha S. Sanagala Andrew Nicolaides Suneet K. Gupta Vijaya K. Koppula Luca Saba Sushant Agarwal Amer M. Johri Manudeep S. Kalra Jasjit S. Suri
author_sort	Skandha S. Sanagala
title	Ten Fast Transfer Learning Models for Carotid Ultrasound Plaque Tissue Characterization in Augmentation Framework Embedded with Heatmaps for Stroke Risk Stratification
title_short	Ten Fast Transfer Learning Models for Carotid Ultrasound Plaque Tissue Characterization in Augmentation Framework Embedded with Heatmaps for Stroke Risk Stratification
title_full	Ten Fast Transfer Learning Models for Carotid Ultrasound Plaque Tissue Characterization in Augmentation Framework Embedded with Heatmaps for Stroke Risk Stratification
title_fullStr	Ten Fast Transfer Learning Models for Carotid Ultrasound Plaque Tissue Characterization in Augmentation Framework Embedded with Heatmaps for Stroke Risk Stratification
title_full_unstemmed	Ten Fast Transfer Learning Models for Carotid Ultrasound Plaque Tissue Characterization in Augmentation Framework Embedded with Heatmaps for Stroke Risk Stratification
title_sort	ten fast transfer learning models for carotid ultrasound plaque tissue characterization in augmentation framework embedded with heatmaps for stroke risk stratification
publisher	MDPI AG
publishDate	2021
url	https://doaj.org/article/328b577cc70340eaa8d8697fe1e80bbb
work_keys_str_mv	AT skandhassanagala tenfasttransferlearningmodelsforcarotidultrasoundplaquetissuecharacterizationinaugmentationframeworkembeddedwithheatmapsforstrokeriskstratification AT andrewnicolaides tenfasttransferlearningmodelsforcarotidultrasoundplaquetissuecharacterizationinaugmentationframeworkembeddedwithheatmapsforstrokeriskstratification AT suneetkgupta tenfasttransferlearningmodelsforcarotidultrasoundplaquetissuecharacterizationinaugmentationframeworkembeddedwithheatmapsforstrokeriskstratification AT vijayakkoppula tenfasttransferlearningmodelsforcarotidultrasoundplaquetissuecharacterizationinaugmentationframeworkembeddedwithheatmapsforstrokeriskstratification AT lucasaba tenfasttransferlearningmodelsforcarotidultrasoundplaquetissuecharacterizationinaugmentationframeworkembeddedwithheatmapsforstrokeriskstratification AT sushantagarwal tenfasttransferlearningmodelsforcarotidultrasoundplaquetissuecharacterizationinaugmentationframeworkembeddedwithheatmapsforstrokeriskstratification AT amermjohri tenfasttransferlearningmodelsforcarotidultrasoundplaquetissuecharacterizationinaugmentationframeworkembeddedwithheatmapsforstrokeriskstratification AT manudeepskalra tenfasttransferlearningmodelsforcarotidultrasoundplaquetissuecharacterizationinaugmentationframeworkembeddedwithheatmapsforstrokeriskstratification AT jasjitssuri tenfasttransferlearningmodelsforcarotidultrasoundplaquetissuecharacterizationinaugmentationframeworkembeddedwithheatmapsforstrokeriskstratification
_version_	1718412461623738368

Ten Fast Transfer Learning Models for Carotid Ultrasound Plaque Tissue Characterization in Augmentation Framework Embedded with Heatmaps for Stroke Risk Stratification

Ejemplares similares