Using deep learning to screen OCTA images for hypertension to reduce the risk of serious complications

BackgroundAs a disease with high global incidence, hypertension is known to cause systemic vasculopathy. Ophthalmic vessels are the only vascular structures that can be directly observed in vivo in a non-invasive manner. We aim to investigate the changes in ocular microvessels in hypertension using...

Full description

Saved in:
Bibliographic Details
Main Authors: Yiheng Ding, Ziqiang Wei, Chaoyun Wang, Xinyue Li, Bingbing Li, Xueting Liu, Zhijie Fu, Hongwei Mo, Hong Zhang
Format: Article
Language:English
Published: Frontiers Media S.A. 2025-07-01
Series:Frontiers in Cell and Developmental Biology
Subjects:
Online Access:https://www.frontiersin.org/articles/10.3389/fcell.2025.1581785/full
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849320561322229760
author Yiheng Ding
Ziqiang Wei
Chaoyun Wang
Xinyue Li
Bingbing Li
Xueting Liu
Zhijie Fu
Hongwei Mo
Hong Zhang
author_facet Yiheng Ding
Ziqiang Wei
Chaoyun Wang
Xinyue Li
Bingbing Li
Xueting Liu
Zhijie Fu
Hongwei Mo
Hong Zhang
author_sort Yiheng Ding
collection DOAJ
description BackgroundAs a disease with high global incidence, hypertension is known to cause systemic vasculopathy. Ophthalmic vessels are the only vascular structures that can be directly observed in vivo in a non-invasive manner. We aim to investigate the changes in ocular microvessels in hypertension using deep learning on optical coherence tomography angiography (OCTA) images.MethodsThe convolutional neural network architecture Xception and multi-Swin transformer were used to screen 422 OCTA images (252 from 136 hypertension subjects; 170 from 85 healthy subjects) for hypertension. Moreover, the separability of the OCTA images based on high-dimensional feature angles was analyzed to better understand how deep learning models distinguish such images with class activation mapping.ResultsUnder Xception, the overall average accuracy of 5-fold cross-validation was 76.05% and sensitivity was 85.52%. In contrast, the Swin transformer showed single-model (macular), single-model (optic disk), and multimodel average accuracies of 82.25%, 74.936%, and 85.06%, respectively, for predicting hypertension.ConclusionThe changes caused by hypertension on the fundus vessels can be observed more accurately and efficiently using OCTA image features through deep learning. These results are expected to assist with screening of hypertension and reducing the risk of its severe complications.Trial RegistrationChiCTR, ChiCTR2000041330. Registered 23 December 2020, https://www.chictr.org.cn/ChiCTR2000041330.
format Article
id doaj-art-60f8c9e4f13247d396f5fbcdc52a7c2b
institution Kabale University
issn 2296-634X
language English
publishDate 2025-07-01
publisher Frontiers Media S.A.
record_format Article
series Frontiers in Cell and Developmental Biology
spelling doaj-art-60f8c9e4f13247d396f5fbcdc52a7c2b2025-08-20T03:50:01ZengFrontiers Media S.A.Frontiers in Cell and Developmental Biology2296-634X2025-07-011310.3389/fcell.2025.15817851581785Using deep learning to screen OCTA images for hypertension to reduce the risk of serious complicationsYiheng Ding0Ziqiang Wei1Chaoyun Wang2Xinyue Li3Bingbing Li4Xueting Liu5Zhijie Fu6Hongwei Mo7Hong Zhang8Eye Hospital, The First Affiliated Hospital of Harbin Medical University, Harbin, ChinaSchool of Intelligent Science and Engineering, Harbin Engineering University, Harbin, ChinaSchool of Intelligent Science and Engineering, Harbin Engineering University, Harbin, ChinaEye Hospital, The First Affiliated Hospital of Harbin Medical University, Harbin, ChinaSchool of Intelligent Science and Engineering, Harbin Engineering University, Harbin, ChinaEye Hospital, The First Affiliated Hospital of Harbin Medical University, Harbin, ChinaSchool of Intelligent Science and Engineering, Harbin Engineering University, Harbin, ChinaSchool of Intelligent Science and Engineering, Harbin Engineering University, Harbin, ChinaEye Hospital, The First Affiliated Hospital of Harbin Medical University, Harbin, ChinaBackgroundAs a disease with high global incidence, hypertension is known to cause systemic vasculopathy. Ophthalmic vessels are the only vascular structures that can be directly observed in vivo in a non-invasive manner. We aim to investigate the changes in ocular microvessels in hypertension using deep learning on optical coherence tomography angiography (OCTA) images.MethodsThe convolutional neural network architecture Xception and multi-Swin transformer were used to screen 422 OCTA images (252 from 136 hypertension subjects; 170 from 85 healthy subjects) for hypertension. Moreover, the separability of the OCTA images based on high-dimensional feature angles was analyzed to better understand how deep learning models distinguish such images with class activation mapping.ResultsUnder Xception, the overall average accuracy of 5-fold cross-validation was 76.05% and sensitivity was 85.52%. In contrast, the Swin transformer showed single-model (macular), single-model (optic disk), and multimodel average accuracies of 82.25%, 74.936%, and 85.06%, respectively, for predicting hypertension.ConclusionThe changes caused by hypertension on the fundus vessels can be observed more accurately and efficiently using OCTA image features through deep learning. These results are expected to assist with screening of hypertension and reducing the risk of its severe complications.Trial RegistrationChiCTR, ChiCTR2000041330. Registered 23 December 2020, https://www.chictr.org.cn/ChiCTR2000041330.https://www.frontiersin.org/articles/10.3389/fcell.2025.1581785/fulloptical coherence tomography angiographyhypertensiondeep learningconvolutional neural networkmulti-Swin transformer
spellingShingle Yiheng Ding
Ziqiang Wei
Chaoyun Wang
Xinyue Li
Bingbing Li
Xueting Liu
Zhijie Fu
Hongwei Mo
Hong Zhang
Using deep learning to screen OCTA images for hypertension to reduce the risk of serious complications
Frontiers in Cell and Developmental Biology
optical coherence tomography angiography
hypertension
deep learning
convolutional neural network
multi-Swin transformer
title Using deep learning to screen OCTA images for hypertension to reduce the risk of serious complications
title_full Using deep learning to screen OCTA images for hypertension to reduce the risk of serious complications
title_fullStr Using deep learning to screen OCTA images for hypertension to reduce the risk of serious complications
title_full_unstemmed Using deep learning to screen OCTA images for hypertension to reduce the risk of serious complications
title_short Using deep learning to screen OCTA images for hypertension to reduce the risk of serious complications
title_sort using deep learning to screen octa images for hypertension to reduce the risk of serious complications
topic optical coherence tomography angiography
hypertension
deep learning
convolutional neural network
multi-Swin transformer
url https://www.frontiersin.org/articles/10.3389/fcell.2025.1581785/full
work_keys_str_mv AT yihengding usingdeeplearningtoscreenoctaimagesforhypertensiontoreducetheriskofseriouscomplications
AT ziqiangwei usingdeeplearningtoscreenoctaimagesforhypertensiontoreducetheriskofseriouscomplications
AT chaoyunwang usingdeeplearningtoscreenoctaimagesforhypertensiontoreducetheriskofseriouscomplications
AT xinyueli usingdeeplearningtoscreenoctaimagesforhypertensiontoreducetheriskofseriouscomplications
AT bingbingli usingdeeplearningtoscreenoctaimagesforhypertensiontoreducetheriskofseriouscomplications
AT xuetingliu usingdeeplearningtoscreenoctaimagesforhypertensiontoreducetheriskofseriouscomplications
AT zhijiefu usingdeeplearningtoscreenoctaimagesforhypertensiontoreducetheriskofseriouscomplications
AT hongweimo usingdeeplearningtoscreenoctaimagesforhypertensiontoreducetheriskofseriouscomplications
AT hongzhang usingdeeplearningtoscreenoctaimagesforhypertensiontoreducetheriskofseriouscomplications