A Deep Learning Architecture for Vascular Area Measurement in Fundus Images

Kanae Fukutsu; Michiyuki Saito; Kousuke Noda; Miyuki Murata; Satoru Kase; Ryosuke Shiba; Naoki Isogai; Yoshikazu Asano; Nagisa Hanawa; Mitsuru Dohke; Manabu Kase; Susumu Ishida

doi:10.1016/j.xops.2021.100004

A Deep Learning Architecture for Vascular Area Measurement in Fundus Images

Ophthalmol Sci. 2021 Feb 23;1(1):100004. doi: 10.1016/j.xops.2021.100004. eCollection 2021 Mar.

Authors

Kanae Fukutsu¹, Michiyuki Saito¹, Kousuke Noda^{1

2}, Miyuki Murata^{1

2}, Satoru Kase¹, Ryosuke Shiba³, Naoki Isogai³, Yoshikazu Asano⁴, Nagisa Hanawa⁴, Mitsuru Dohke⁴, Manabu Kase⁵, Susumu Ishida^{1

2}

Affiliations

¹ Department of Ophthalmology, Faculty of Medicine and Graduate School of Medicine, Hokkaido University, Sapporo, Japan.
² Department of Ocular Circulation and Metabolism, Hokkaido University, Sapporo, Japan.
³ NIDEK Co., Ltd., Gamagori, Japan.
⁴ Keijinkai Maruyama Clinic, Sapporo, Japan.
⁵ Department of Ophthalmology, Teine Keijinkai Hospital, Sapporo, Japan.

Abstract

Purpose: To develop a novel evaluation system for retinal vessel alterations caused by hypertension using a deep learning algorithm.

Design: Retrospective study.

Participants: Fundus photographs (n = 10 571) of health-check participants (n = 5598).

Methods: The participants were analyzed using a fully automatic architecture assisted by a deep learning system, and the total area of retinal arterioles and venules was assessed separately. The retinal vessels were extracted automatically from each photograph and categorized as arterioles or venules. Subsequently, the total arteriolar area (AA) and total venular area (VA) were measured. The correlations among AA, VA, age, systolic blood pressure (SBP), and diastolic blood pressure were analyzed. Six ophthalmologists manually evaluated the arteriovenous ratio (AVR) in fundus images (n = 102), and the correlation between the SBP and AVR was evaluated manually.

Main outcome measures: Total arteriolar area and VA.

Results: The deep learning algorithm demonstrated favorable properties of vessel segmentation and arteriovenous classification, comparable with pre-existing techniques. Using the algorithm, a significant positive correlation was found between AA and VA. Both AA and VA demonstrated negative correlations with age and blood pressure. Furthermore, the SBP showed a higher negative correlation with AA measured by the algorithm than with AVR.

Conclusions: The current data demonstrated that the retinal vascular area measured with the deep learning system could be a novel index of hypertension-related vascular changes.

Keywords: AA, total arteriolar area; AUC, area under the receiver operating characteristic curve; AVR, arteriovenous ratio; Arteriosclerosis; BP, blood pressure; DBP, diastolic blood pressure; DRIVE, Digital Retinal Images for Vessel Extraction; Deep learning system; FN, false-negative; FP, false-positive; FPa, FP arterioles; FPv, FP venules; Hypertensive retinopathy; Imaging; MISCa, misclassification rates of arterioles; MISCv, misclassification rates of venules; RGB, red-green-blue; Retinal arteriolar narrowing; SBP, systolic blood pressure; TN, true-negative; TP, true-positive; TPa, TP arterioles; TPv, TP venules; VA, total venular area.