Factors in Color Fundus Photographs That Can Be Used by Humans to Determine Sex of Individuals

Abstract

Purpose: Artificial intelligence (AI) can identify the sex of an individual from color fundus photographs (CFPs). However, the mechanism(s) involved in this identification has not been determined. This study was conducted to determine the information in CFPs that can be used to determine the sex of an individual. Methods:Prospective observational cross-sectional study of 112 eyes of 112 healthy volunteers. The following characteristics of CFPs were analyzed: the color of peripapillary area expressed by the mean values of red, green, and blue intensities, and the tessellation expressed by the tessellation fundus index (TFI). The optic disc ovality ratio, papillomacular angle, retinal artery trajectory, and retinal vessel angles were also quantified. Their differences between the sexes were assessed by Mann-Whitney U tests. Regularized binomial logistic regression was used to select the decisive factors. In addition, its discriminative performance was evaluated through the leave-one-out cross validation. Results:The mean age of 76 men and 36 women was 25.8 years. The regularized binomial logistic regression delivered the optimal model for sex selected variables of peripapillary temporal green and blue intensities, temporal TFI, supratemporal TFI, optic disc ovality ratio, artery trajectory, and supratemporal retinal artery angle. With this approach, the discrimination accuracy rate was 77.9%. Conclusions:Human-assessed characteristics of CFPs are useful in investigating the new theme proposed by AI, the sex of an individual.Translational Relevance: This is the first report to approach the thinking process of AI by humans and can be a new approach to medical AI research.

Sex judgment using color fundus parameters in elementary school students

“…We recently constructed a model to discriminate sex from the clinical parameters of CFP known by humans, such as angle or trajectory of retinal vessels [18][19][20][21][22][23], location and shape of the optic disc [24,25], and color intensity in the peripapillary area [26][27][28][29][30][31][32], albeit using the binomial regression with L2 regularization (Ridge regression) [33,34], in adults [35]. As a result, this allowed us to partially, at least, reproduce sex discrimination by deep learning with an AUC value of 77.9%.…”

Factors in Color Fundus Photographs That Can Be Used by Humans to Determine Sex of Individuals

“…We were intrigued by the findings of Yamashita et al in their recent paper "Factors in Color Fundus Photographs That Can Be Used by Humans to Determine Sex of Individuals." 1 As mentioned by the authors, the publication by Poplin et al in 2018 captivated the interest of many ophthalmologists when the authors proposed a deep neural network for sex differentiation on color fundus photography (CFP) with an area under the curve (AUC) of 97%. 2 However, to date, the results have yet to be replicated, the neural network model was not shared for independent validation nor explained in detail, and the published saliency maps did not highlight any specific features except for the fovea and the optic disc.…”

“…The features identified by the authors were optic disc ovality ratio, papillomacular angle, retinal artery trajectory, and retinal vessel angles, the mean red, green, and blue colors of the peripapillary area expressed by a tessellation index. 1 In order to evaluate AI-based approaches for sex determination on CFP, we have performed meticulous analyses, including the reproducibility of the results from Poplin et al, the evaluation of a deep learning (DL) approach for assessing the informative relevance of previously known and human interpretable features compared to unknown/unstructured image information, and, finally, assessment of human performance against that of AI.…”

“…The features identified by the authors were optic disc ovality ratio, papillomacular angle, retinal artery trajectory, and retinal vessel angles, the mean red, green, and blue colors of the peripapillary area expressed by a tessellation index. 1 …”

Author Response: Factors in Color Fundus Photographs That Can Be Used by Humans to Determine Sex of Individuals

“…We thank Dr Dieck and co-authors 1 for the comments on our paper. 2 They performed an important experiment. In that, ophthalmologists were told about some features of the difference in the fundus image of men and women in advance and judged the sex of each eye from her/his fundus photograph.…”