Abstract
Purpose :
Recently, artificial intelligence estimated the sex using only fundus photographs. And we reported that sex can be distinguished by known factors obtained from color fundus photography in young healthy eyes. However, it’s not clear when the sex difference of fundus parameters occur. Therefore, the purpose of this study was to investigate gender determination by fundus parameters using binominal logistic regression in elementary school students.
Methods :
A prospective observational cross-sectional study was performed in 119 right eyes of elementary school students (8 to 9 years). Using the color fundus photograph, Tessellation fundus index (TFI) was calculated by TFI=R/(R+G+B) using the mean value of red-green-blue intensity in the eight locations around the optic disc. Optic disc ovality ratio, papillo-macular angle, retinal artery trajectory, retinal vessel were quantified by our previous reports. The binomial logistic regression was used to predict the gender by these fundus parameters. Sex difference of the parameters were assessed by Mann-Whitney U test.
Results :
59 boys and 60 girls. In the backward selection method, 37 parameters were selected, and the predicted correct answer rate was 89.1%. Ovality ratio in girls was significantly smaller than that in boys. Nasal green and blue intensity of girls were significantly higher than that of boys.
Conclusions :
It was already possible to estimate boys and girls by color fundus parameters in elementary school students.
This is a 2020 ARVO Annual Meeting abstract.