Prediction of age and brachial-ankle pulse-wave velocity using ultra-wide-field pseudo-color images by deep learning

Abstract

This study examined whether age and brachial-ankle pulse-wave velocity (baPWV) can be predicted with ultra-wide-field pseudo-color (UWPC) images using deep learning (DL). We examined 170 UWPC images of both eyes of 85 participants (40 men and 45 women, mean age: 57.5 ± 20.9 years). Three types of images were included (total, central, and peripheral) and analyzed by k-fold cross-validation (k = 5) using Visual Geometry Group-16. After bias was eliminated using the generalized linear mixed model, the standard regression coefficients (SRCs) between actual age and baPWV and predicted age and baPWV from the UWPC images by the neural network were calculated, and the prediction accuracies of the DL model for age and baPWV were examined. The SRC between actual age and predicted age by the neural network was 0.833 for all images, 0.818 for central images, and 0.649 for peripheral images (all P < 0.001) and between the actual baPWV and the predicted baPWV was 0.390 for total images, 0.419 for central images, and 0.312 for peripheral images (all P < 0.001). These results show the potential prediction capability of DL for age and vascular aging and could be useful for disease prevention and early treatment.

Deep learning of fundus and optical coherence tomography images enables identification of diverse genetic and environmental factors associated with eye aging

“…More recently Nagasato et al used transfer learning to train a VGG16 architecture on a far smaller sample size (n=85) using ultra-wide-field pseudo-color eye fundus images collected from participants aged 57.5±20.9 years to predict chronological age. However, they did not report their R-Squared value, only the slope of the regression coefficient between their predictor and chronological age (standard regression coefficient=0.833) 6 .…”

“…Accelerated eye aging is defined as the difference between eye age and age. Age has for example already been predicted from fundus 5,6 , iris [7][8][9] and eye corner images 10 . Identifying the genetic and non-genetic factors underlying accelerated eye aging however remains an elusive question.…”

“…For example, our InceptionV3 model slightly outperformed their model (R 2 =75.2±0.1%), whereas our InceptionResNetV2 model significantly underperformed (R 2 =69.1±0.2%).More recently Nagasato et al used transfer learning to train a VGG16 architecture on a far smaller sample size (n=85) using ultra-wide-field pseudo-color eye fundus images collected from participants aged 57.5±20.9 years to predict chronological age. However, they did not report their R-Squared value, only the slope of the regression coefficient between their predictor and chronological age (standard regression coefficient=0.833)6 .Bobrov et al trained CNNs on 8,414 eye corner images collected from participants aged 20-80 and predicted chronological age with a R 2 value of 90.25% (estimated from the Pearson correlation) and a MAE of 2.3 years 10 .…”

The Correlation of Atherosclerosis and Triglyceride Glucose Index: A Secondary Analysis of a National Cross-Sectional Study of Japanese

“…The brachial ankle pulse wave velocity (baPWV) is a candidate measure to estimate arteriosclerosis progression quantitatively noninvasively [2]. In clinical practice, the baPWV values indicate the degree of systemic arteriosclerosis and vascular disorders [3,4].…”

“…Therefore, the positive association between TyG index and baPWV was robust.The strengths of this study include: (1) our study had stricter cohort criteria for inclusion by excluding patients with symptomatic coronary artery disease (CAD) and included a broad age range (24-84 years), not merely older persons, as well as ABI<0.95 (2). This study is the first to report the relationship between the TyG index and baPWV in Japan (3). We analyse the TyG index as both continuous and categorical variables to reduce the result contingency and increase the result robustness.…”