Ocular Axial Length Prediction Based on Visual Interpretation of Retinal Fundus Images via Deep Neural Network

Yeonwoo Jeong; Boram Lee; Jae Ho Han; Jaeryung Oh

doi:10.1109/JSTQE.2020.3038845

Ocular Axial Length Prediction Based on Visual Interpretation of Retinal Fundus Images via Deep Neural Network

Yeonwoo Jeong, Boram Lee, Jae Ho Han, Jaeryung Oh

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

14 Citations (Scopus)

Abstract

Ocular axial length (AL) is an important property of eyes used for determining their health prior to surgery. Estimation of AL is also crucial while making artificial lenses to replace impaired natural lenses. However, accurate measurement of AL requires a costly and bulky benchtop optical system. The complex structural features of eyes can be captured by fundus images, which can be easily captured nowadays with portable cameras. Here, we suggest a deep learning method for predicting AL based on fundus images with evidence of decision. This visual interpretation of predictions is achieved by post-processing, separated from the training process, to ensure that the architecture can be freely designed. Through the visualization technique, discriminative regions on input images can be localized to demonstrate specific areas of interest for predictions. In the experiments, we found a significant relationship between the fundus images and AL with achieving a coefficient of determination (R2) of 0.67 and accuracy of 90%, within an error margin of $ \pm 1$ mm. Furthermore, visual evidence proves that the network uses consistent regions for predicting AL. The visual results of this study also point to a link between AL and biological structure of eyes, which paves the way for future research.

Original language	English
Article number	9264730
Journal	IEEE Journal of Selected Topics in Quantum Electronics
Volume	27
Issue number	4
DOIs	https://doi.org/10.1109/JSTQE.2020.3038845
Publication status	Published - 2021 Jul 1

Bibliographical note

Funding Information:
Manuscript received September 18, 2020; revised November 4, 2020; accepted November 14, 2020. Date of publication November 19, 2020; date of current version December 3, 2020. This work was supported in part by the MSIT (Ministry of Science and ICT), Korea, under the ITRC (Information Technology Research Center) support program (IITP-2020-2016-0-00464) supervised by the IITP (Institute for Information & Communications Technology Planning & Evaluation), in part by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (NRF-2020R1A4A1018309), in part by Korea University Future Research Grant (FRG), and in part by the Ministry of Trade, Industry & Energy (MOTIE, Korea) under the Industrial Technology Innovation (10063364). (Corresponding author: Jae-Ho Han.) Yeonwoo Jeong is with the Department of Brain and Cognitive Engineering, Korea University, Seoul 02841, South Korea (e-mail: forresearch4220@ gmail.com).

Publisher Copyright:
© 1995-2012 IEEE.

Keywords

Artificial neural networks
biomedical imaging
machine learning
medical diagnosis
regression analysis

ASJC Scopus subject areas

Atomic and Molecular Physics, and Optics
Electrical and Electronic Engineering

Access to Document

10.1109/JSTQE.2020.3038845

Cite this

@article{32a764e3a67d4e06bcc40fd50af560e0,

title = "Ocular Axial Length Prediction Based on Visual Interpretation of Retinal Fundus Images via Deep Neural Network",

abstract = "Ocular axial length (AL) is an important property of eyes used for determining their health prior to surgery. Estimation of AL is also crucial while making artificial lenses to replace impaired natural lenses. However, accurate measurement of AL requires a costly and bulky benchtop optical system. The complex structural features of eyes can be captured by fundus images, which can be easily captured nowadays with portable cameras. Here, we suggest a deep learning method for predicting AL based on fundus images with evidence of decision. This visual interpretation of predictions is achieved by post-processing, separated from the training process, to ensure that the architecture can be freely designed. Through the visualization technique, discriminative regions on input images can be localized to demonstrate specific areas of interest for predictions. In the experiments, we found a significant relationship between the fundus images and AL with achieving a coefficient of determination (R2) of 0.67 and accuracy of 90%, within an error margin of $ \pm 1$ mm. Furthermore, visual evidence proves that the network uses consistent regions for predicting AL. The visual results of this study also point to a link between AL and biological structure of eyes, which paves the way for future research. ",

keywords = "Artificial neural networks, biomedical imaging, machine learning, medical diagnosis, regression analysis",

author = "Yeonwoo Jeong and Boram Lee and Han, {Jae Ho} and Jaeryung Oh",

note = "Funding Information: Manuscript received September 18, 2020; revised November 4, 2020; accepted November 14, 2020. Date of publication November 19, 2020; date of current version December 3, 2020. This work was supported in part by the MSIT (Ministry of Science and ICT), Korea, under the ITRC (Information Technology Research Center) support program (IITP-2020-2016-0-00464) supervised by the IITP (Institute for Information & Communications Technology Planning & Evaluation), in part by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (NRF-2020R1A4A1018309), in part by Korea University Future Research Grant (FRG), and in part by the Ministry of Trade, Industry & Energy (MOTIE, Korea) under the Industrial Technology Innovation (10063364). (Corresponding author: Jae-Ho Han.) Yeonwoo Jeong is with the Department of Brain and Cognitive Engineering, Korea University, Seoul 02841, South Korea (e-mail: forresearch4220@ gmail.com). Publisher Copyright: {\textcopyright} 1995-2012 IEEE.",

year = "2021",

month = jul,

day = "1",

doi = "10.1109/JSTQE.2020.3038845",

language = "English",

volume = "27",

journal = "IEEE Journal of Selected Topics in Quantum Electronics",

issn = "1077-260X",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "4",

}

TY - JOUR

T1 - Ocular Axial Length Prediction Based on Visual Interpretation of Retinal Fundus Images via Deep Neural Network

AU - Jeong, Yeonwoo

AU - Lee, Boram

AU - Han, Jae Ho

AU - Oh, Jaeryung

N1 - Funding Information: Manuscript received September 18, 2020; revised November 4, 2020; accepted November 14, 2020. Date of publication November 19, 2020; date of current version December 3, 2020. This work was supported in part by the MSIT (Ministry of Science and ICT), Korea, under the ITRC (Information Technology Research Center) support program (IITP-2020-2016-0-00464) supervised by the IITP (Institute for Information & Communications Technology Planning & Evaluation), in part by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (NRF-2020R1A4A1018309), in part by Korea University Future Research Grant (FRG), and in part by the Ministry of Trade, Industry & Energy (MOTIE, Korea) under the Industrial Technology Innovation (10063364). (Corresponding author: Jae-Ho Han.) Yeonwoo Jeong is with the Department of Brain and Cognitive Engineering, Korea University, Seoul 02841, South Korea (e-mail: forresearch4220@ gmail.com). Publisher Copyright: © 1995-2012 IEEE.

PY - 2021/7/1

Y1 - 2021/7/1

N2 - Ocular axial length (AL) is an important property of eyes used for determining their health prior to surgery. Estimation of AL is also crucial while making artificial lenses to replace impaired natural lenses. However, accurate measurement of AL requires a costly and bulky benchtop optical system. The complex structural features of eyes can be captured by fundus images, which can be easily captured nowadays with portable cameras. Here, we suggest a deep learning method for predicting AL based on fundus images with evidence of decision. This visual interpretation of predictions is achieved by post-processing, separated from the training process, to ensure that the architecture can be freely designed. Through the visualization technique, discriminative regions on input images can be localized to demonstrate specific areas of interest for predictions. In the experiments, we found a significant relationship between the fundus images and AL with achieving a coefficient of determination (R2) of 0.67 and accuracy of 90%, within an error margin of $ \pm 1$ mm. Furthermore, visual evidence proves that the network uses consistent regions for predicting AL. The visual results of this study also point to a link between AL and biological structure of eyes, which paves the way for future research.

AB - Ocular axial length (AL) is an important property of eyes used for determining their health prior to surgery. Estimation of AL is also crucial while making artificial lenses to replace impaired natural lenses. However, accurate measurement of AL requires a costly and bulky benchtop optical system. The complex structural features of eyes can be captured by fundus images, which can be easily captured nowadays with portable cameras. Here, we suggest a deep learning method for predicting AL based on fundus images with evidence of decision. This visual interpretation of predictions is achieved by post-processing, separated from the training process, to ensure that the architecture can be freely designed. Through the visualization technique, discriminative regions on input images can be localized to demonstrate specific areas of interest for predictions. In the experiments, we found a significant relationship between the fundus images and AL with achieving a coefficient of determination (R2) of 0.67 and accuracy of 90%, within an error margin of $ \pm 1$ mm. Furthermore, visual evidence proves that the network uses consistent regions for predicting AL. The visual results of this study also point to a link between AL and biological structure of eyes, which paves the way for future research.

KW - Artificial neural networks

KW - biomedical imaging

KW - machine learning

KW - medical diagnosis

KW - regression analysis

UR - http://www.scopus.com/inward/record.url?scp=85096851953&partnerID=8YFLogxK

U2 - 10.1109/JSTQE.2020.3038845

DO - 10.1109/JSTQE.2020.3038845

M3 - Article

AN - SCOPUS:85096851953

SN - 1077-260X

VL - 27

JO - IEEE Journal of Selected Topics in Quantum Electronics

JF - IEEE Journal of Selected Topics in Quantum Electronics

IS - 4

M1 - 9264730

ER -

Ocular Axial Length Prediction Based on Visual Interpretation of Retinal Fundus Images via Deep Neural Network

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this