Unveiling Associations Between Retinal Microvascular Architecture and Phenotypes Across Thousands of Healthy Subjects

Michal Shapira, Smadar Shilo, Yeela Talmor‐Barkan, Yaron Aviv, Yotam Reisner, Anastasia Godneva, Adina Weinberger, Alon Skaat, Anat Loewenstein, Eran Segal, Hagai RossmanmedRxiv 2024 · DOI

Abstract

Fundus imaging has emerged as a powerful tool, allowing high-resolution and non-invasive visualization of the fundus microvasculature. Recent advancements in artificial intelligence (AI) have enabled the quantification of fundus microvascular characteristics, paving the way for the development of predictive markers for systemic conditions. In this study, we characterize the fundus microvasculature among 8,467 healthy individuals, aged 40-70 years. We employ a fully automated segmentation tool to calculate 12 distinct vessel measures, including tortuosity density, vessel density, average width, fractal dimension, distance tortuosity, and curvature tortuosity for both arteries and veins. Our analysis reveals numerous significant correlations, including connections between fundus arterial measures and cardiovascular measurements, lipid profile, sleep apnea indices, glycemic profile, body composition, and smoking status. This comprehensive analysis contributes to the growing field of oculomics and highlights the potential of fundus imaging for earlier detection and monitoring of systemic diseases.

Introduction: The Promise of Oculomics

The field of oculomics—using the eye as a non-invasive window into the body’s systemic health—has long been hampered by a lack of scale and standardization. While it is well-established that conditions like diabetes and hypertension manifest in the retina, previous research has often been limited by small sample sizes or labor-intensive manual measurements. To utilize the eye as a predictive tool for the general population, a robust understanding of "normal" microvascular architecture is required. This study addresses this gap by characterizing the retinal microvasculature in a massive, healthy cohort, providing a baseline for how these vessels change with age and sex.

Methodology: Automated Microvascular Quantification

The researchers utilized AutoMorph, a fully automated, deep-learning-based pipeline for segmenting and quantifying fundus (retinal) images. This high-throughput approach extracted 12 distinct vascular measures (6 for arteries and 6 for veins), allowing for the objective analysis of thousands of images.

Visualizing retinal vascular measures: Examples of high and low fractal dimension, tortuosity density, and vessel density with corresponding segmentation masks.

Key metrics analyzed include:

  • Fractal Dimension: Complexity of the branching pattern.
  • Vessel Density: Fraction of the total area covered by blood vessels.
  • Average Width: Thickness of the vessels.
  • Tortuosity (Density, Distance, and Curvature): Quantification of vessel "twistiness."

Data Sources: The 10K Project Cohort

Data was drawn from the 10K project, a longitudinal study in Israel. The analysis included 8,467 healthy individuals (aged 40–70) after excluding those with pre-existing eye conditions. This dataset is uniquely multimodal, combining fundus images with DXA scans (body composition), Continuous Glucose Monitoring (CGM), sleep tests (WatchPAT), and comprehensive blood panels.

Results: Age, Sex, and Systemic Health Correlations

The Aging Eye and Sex Dimorphism

Retinal architecture changes predictably over time. Artery width and fractal dimension (complexity) decrease with age, indicating a vascular network that becomes "thinner" and less branched. Conversely, vessel tortuosity increases as individuals age. Significant sex differences were also observed; for instance, artery fractal dimension is typically higher in males, while tortuosity and average width are generally higher in females.

Trends in retinal microvascular measures by age and sex, showing correlations and significant differences between males and females.
  • Cardiovascular and Metabolic Markers: Artery vessel density and fractal dimension emerged as robust indicators of systemic health, showing significant negative correlations with BMI, waist-to-hip ratio, and blood pressure. High blood pressure and carotid intima-media thickness were consistently associated with increased vessel tortuosity and decreased vessel width.
  • Sleep Apnea: A novel finding was the strong negative correlation between artery vessel density/fractal dimension and sleep apnea indices (AHI and ODI), suggesting that chronic intermittent hypoxia may lead to microvascular "rarefaction."
  • Smoking Status: Current smokers exhibited a higher artery fractal dimension, potentially representing a pathological angiogenic response to compensate for decreased oxygen-carrying capacity.

Discussion and Future Directions

While this cross-sectional study identifies strong associations, it does not prove causality. The fully automated nature of the pipeline, while efficient, may also include some suboptimal images. However, the consistency of findings—such as the minimal difference between an individual's left and right eye—validates the use of these measures for systemic health assessment.

Future work will focus on moving from association to prediction. By following this cohort longitudinally, researchers hope to determine if these retinal biomarkers can predict the onset of metabolic syndrome or cardiovascular events years before traditional clinical symptoms appear.

Summary updated

Abstract

Abstract Fundus imaging has emerged as a powerful tool, allowing high-resolution and non-invasive visualization of the fundus microvasculature. Recent advancements in artificial intelligence (AI) have enabled the quantification of fundus microvascular characteristics, paving the way for the development of predictive markers for systemic conditions. In this study, we characterize the fundus microvasculature among 8467 healthy individuals, aged 40-70 years. We employ a fully automated segmentation tool to calculate 12 distinct vessel measures, including tortuosity density, vessel density, average width, fractal dimension, distance tortuosity, and curvature tortuosity for both arteries and veins. We provide reference values for these measures by age and sex, and explore the associations between these fundus vessel measures and various clinical features. Our analysis reveals numerous significant correlations, including connections between fundus arterial measures and cardiovascular measurements, lipid profile, sleep apnea indices, glycemic profile, body composition, and smoking status. This comprehensive analysis of fundus microvasculature anatomy in a large healthy population contributes to the growing field of oculomics and highlights the potential of fundus imaging for earlier detection and monitoring of systemic diseases such as metabolic syndrome, cardiovascular diseases and sleep disorders.
BibTeX
@misc{shapira2024unveiling,
  author = {Michal Shapira and Smadar Shilo and Yeela Talmor‐Barkan and Yaron Aviv and Yotam Reisner and Anastasia Godneva and Adina Weinberger and Alon Skaat and Anat Loewenstein and Eran Segal and Hagai Rossman},
  title = {{Unveiling Associations Between Retinal Microvascular Architecture and Phenotypes Across Thousands of Healthy Subjects}},
  year = {2024},
  doi = {10.1101/2024.04.05.24305164},
  howpublished = {medRxiv},
}
Figure