Early Detection of Dementia through Retinal Imaging and Trustworthy AI

Jinkui Hao; William R.  Kwapong; Ting Shen; Huazhu  Fu; Yanwu Xu; Qinkang  Lu; Shouyue Liu; Jiong Zhang; YONGHUAI LIU; Yifan Zhao; Yalin Zheng; Alejandro F  Frangi; Shuting  Zhang; Hong Qi; Yitian Zhao

doi:10.1038/s41746-024-01292-5

Early Detection of Dementia through Retinal Imaging and Trustworthy AI

Jinkui Hao, William R. Kwapong, Ting Shen, Huazhu Fu, Yanwu Xu, Qinkang Lu, Shouyue Liu, Jiong Zhang, YONGHUAI LIU, Yifan Zhao, Yalin Zheng, Alejandro F Frangi, Shuting Zhang, Hong Qi, Yitian Zhao^*

^*Corresponding author for this work

Computer Science

Research output: Contribution to journal › Article (journal) › peer-review

3 Downloads (Pure)

Abstract

Alzheimer’s disease (AD) is a global healthcare challenge lacking a simple and affordable detection method. We propose a novel deep learning framework, Eye-AD, to detect Early-onset Alzheimer’s Disease (EOAD) and Mild Cognitive Impairment (MCI) using OCTA images of retinal microvasculature and choriocapillaris. Eye-AD employs a multilevel graph representation
to analyze intra- and inter-instance relationships in retinal layers. Using 5,751 OCTA images from 1,671 participants in a multi-centre study, our model demonstrated superior performance in EOAD (internal data: AUC = 0.9355, external data: AUC = 0.9007) and MCI detection (internal data: AUC =0.8630, external data: AUC = 0.8037). Furthermore, we explored the associations between retinal structural biomarkers in OCTA images and EOAD/MCI, and the results align well with the
conclusions drawn from our deep learning interpretability analysis. Our findings provide further evidence that retinal OCTA imaging, coupled with artificial intelligence, will serve as a rapid, non-invasive, and affordable dementia detection.

Original language	English
Article number	294
Pages (from-to)	1-5
Number of pages	15
Journal	npj Digital Medicine
Volume	7
Early online date	20 Oct 2024
DOIs	https://doi.org/10.1038/s41746-024-01292-5
Publication status	Published - 20 Oct 2024

Keywords

Early Detection
Dementia
Alzheimer Disease (AD)
Mild Cognitive Impairment (MCI)
Retinal Imaging
OCTA (Optical Coherence Tomography Angiography)
Microvasculature
Deep Learning
Eye-AD Model
Graph Neural Network (GNN)
Retinal Biomarkers
Artificial Intelligence (AI)
Machine Learning
Diagnostic Screening
Noninvasive Methods
Multicenter Study
Choroidal Layers
Interpretability Analysis

Research Centres

Centre for Intelligent Visual Computing Research
Data and Complex Systems Research Centre

Access to Document

10.1038/s41746-024-01292-5Licence: CC BY-NC-ND

Early detection of dementia through retinal imaging and trustworthy AIFinal published version, 5.51 MBLicence: CC BY-NC-ND

Cite this

@article{ac945b650ccc4636b2eb1705a225d4ce,

title = "Early Detection of Dementia through Retinal Imaging and Trustworthy AI",

abstract = "Alzheimer{\textquoteright}s disease (AD) is a global healthcare challenge lacking a simple and affordable detection method. We propose a novel deep learning framework, Eye-AD, to detect Early-onset Alzheimer{\textquoteright}s Disease (EOAD) and Mild Cognitive Impairment (MCI) using OCTA images of retinal microvasculature and choriocapillaris. Eye-AD employs a multilevel graph representationto analyze intra- and inter-instance relationships in retinal layers. Using 5,751 OCTA images from 1,671 participants in a multi-centre study, our model demonstrated superior performance in EOAD (internal data: AUC = 0.9355, external data: AUC = 0.9007) and MCI detection (internal data: AUC =0.8630, external data: AUC = 0.8037). Furthermore, we explored the associations between retinal structural biomarkers in OCTA images and EOAD/MCI, and the results align well with theconclusions drawn from our deep learning interpretability analysis. Our findings provide further evidence that retinal OCTA imaging, coupled with artificial intelligence, will serve as a rapid, non-invasive, and affordable dementia detection.",

keywords = "Early Detection, Dementia, Alzheimer Disease (AD), Mild Cognitive Impairment (MCI), Retinal Imaging, OCTA (Optical Coherence Tomography Angiography), Microvasculature, Deep Learning, Eye-AD Model, Graph Neural Network (GNN), Retinal Biomarkers, Artificial Intelligence (AI), Machine Learning, Diagnostic Screening, Noninvasive Methods, Multicenter Study, Choroidal Layers, Interpretability Analysis",

author = "Jinkui Hao and Kwapong, {William R.} and Ting Shen and Huazhu Fu and Yanwu Xu and Qinkang Lu and Shouyue Liu and Jiong Zhang and YONGHUAI LIU and Yifan Zhao and Yalin Zheng and Frangi, {Alejandro F} and Shuting Zhang and Hong Qi and Yitian Zhao",

year = "2024",

month = oct,

day = "20",

doi = "10.1038/s41746-024-01292-5",

language = "English",

volume = "7",

pages = "1--5",

journal = "npj Digital Medicine",

issn = "2398-6352",

publisher = "Nature Publishing Group",

}

TY - JOUR

T1 - Early Detection of Dementia through Retinal Imaging and Trustworthy AI

AU - Hao, Jinkui

AU - Kwapong, William R.

AU - Shen, Ting

AU - Fu, Huazhu

AU - Xu, Yanwu

AU - Lu, Qinkang

AU - Liu, Shouyue

AU - Zhang, Jiong

AU - LIU, YONGHUAI

AU - Zhao, Yifan

AU - Zheng, Yalin

AU - Frangi, Alejandro F

AU - Zhang, Shuting

AU - Qi, Hong

AU - Zhao, Yitian

PY - 2024/10/20

Y1 - 2024/10/20

N2 - Alzheimer’s disease (AD) is a global healthcare challenge lacking a simple and affordable detection method. We propose a novel deep learning framework, Eye-AD, to detect Early-onset Alzheimer’s Disease (EOAD) and Mild Cognitive Impairment (MCI) using OCTA images of retinal microvasculature and choriocapillaris. Eye-AD employs a multilevel graph representationto analyze intra- and inter-instance relationships in retinal layers. Using 5,751 OCTA images from 1,671 participants in a multi-centre study, our model demonstrated superior performance in EOAD (internal data: AUC = 0.9355, external data: AUC = 0.9007) and MCI detection (internal data: AUC =0.8630, external data: AUC = 0.8037). Furthermore, we explored the associations between retinal structural biomarkers in OCTA images and EOAD/MCI, and the results align well with theconclusions drawn from our deep learning interpretability analysis. Our findings provide further evidence that retinal OCTA imaging, coupled with artificial intelligence, will serve as a rapid, non-invasive, and affordable dementia detection.

AB - Alzheimer’s disease (AD) is a global healthcare challenge lacking a simple and affordable detection method. We propose a novel deep learning framework, Eye-AD, to detect Early-onset Alzheimer’s Disease (EOAD) and Mild Cognitive Impairment (MCI) using OCTA images of retinal microvasculature and choriocapillaris. Eye-AD employs a multilevel graph representationto analyze intra- and inter-instance relationships in retinal layers. Using 5,751 OCTA images from 1,671 participants in a multi-centre study, our model demonstrated superior performance in EOAD (internal data: AUC = 0.9355, external data: AUC = 0.9007) and MCI detection (internal data: AUC =0.8630, external data: AUC = 0.8037). Furthermore, we explored the associations between retinal structural biomarkers in OCTA images and EOAD/MCI, and the results align well with theconclusions drawn from our deep learning interpretability analysis. Our findings provide further evidence that retinal OCTA imaging, coupled with artificial intelligence, will serve as a rapid, non-invasive, and affordable dementia detection.

KW - Early Detection

KW - Dementia

KW - Alzheimer Disease (AD)

KW - Mild Cognitive Impairment (MCI)

KW - Retinal Imaging

KW - OCTA (Optical Coherence Tomography Angiography)

KW - Microvasculature

KW - Deep Learning

KW - Eye-AD Model

KW - Graph Neural Network (GNN)

KW - Retinal Biomarkers

KW - Artificial Intelligence (AI)

KW - Machine Learning

KW - Diagnostic Screening

KW - Noninvasive Methods

KW - Multicenter Study

KW - Choroidal Layers

KW - Interpretability Analysis

U2 - 10.1038/s41746-024-01292-5

DO - 10.1038/s41746-024-01292-5

M3 - Article (journal)

SN - 2398-6352

VL - 7

SP - 1

EP - 5

JO - npj Digital Medicine

JF - npj Digital Medicine

M1 - 294

ER -

Early Detection of Dementia through Retinal Imaging and Trustworthy AI

Abstract

Keywords

Research Centres

Access to Document

Fingerprint

Cite this