The Eye as a Window to Systemic Health: A Survey of Retinal Imaging from Classical Techniques to Oculomics

TL;DR

This survey traces the evolution of retinal imaging from classical techniques to AI-driven oculomics, arguing that the retina acts as a noninvasive window into systemic health. It surveys imaging modalities, publicly available datasets, and automated vascular morphology methods (notably AutoMorph), and discusses how multi-modal and omics data integration can translate retinal traits into cardiovascular, metabolic, and neurodegenerative risk insights. The authors highlight current gaps in data diversity, integration with genomic/proteomic data, and explainability, and propose a roadmap toward foundation-model–driven, multi-system screening tools for precision medicine. Overall, the work emphasizes the eye as a scalable, interpretable biomarker source with potential to transform preventive care and multi-organ disease management.

Abstract

The unique vascularized anatomy of the human eye, encased in the retina, provides an opportunity to act as a window for human health. The retinal structure assists in assessing the early detection, monitoring of disease progression and intervention for both ocular and non-ocular diseases. The advancement in imaging technology leveraging Artificial Intelligence has seized this opportunity to bridge the gap between the eye and human health. This track paves the way for unveiling systemic health insight from the ocular system and surrogating non-invasive markers for timely intervention and identification. The new frontiers of oculomics in ophthalmology cover both ocular and systemic diseases, and getting more attention to explore them. In this survey paper, we explore the evolution of retinal imaging techniques, the dire need for the integration of AI-driven analysis, and the shift of retinal imaging from classical techniques to oculomics. We also discuss some hurdles that may be faced in the progression of oculomics, highlighting the research gaps and future directions.

Figures (7)

• Figure 1: Structural organization of the human eye at multiple anatomical levels. (a) External anatomy of the eye showing the pupil (central opening), iris (colored ring), and sclera (white outer coat that maintains ocular shape and protection). (b) Cross-sectional schematic illustrating the major internal structures including the cornea, iris, lens, vitreous humor, sclera, choroid, retina, blood vessels, and optic nerve. This orientation highlights the key optical and supportive components of the eye. (c) Retinal microstructure depicting the layered neuronal architecture, including photoreceptors (rods and cones), bipolar, horizontal, and amacrine cells, ganglion cells, and the pigment epithelium.
• Figure 2: The schematic illustrates the spectrum of retinal imaging technologies. Modalities shown on a pink background (CFP, OCT, FA, OCTA) represent the core, widely adopted clinical tools that underpin current diagnostic practice. Modalities grouped on a white background (WFDI, HFI, AO, AO-OCT, AO-SLO, cSLO) reflect advanced or emerging approaches that extend the field of view, resolve cellular or biochemical detail, or enhance functional assessment. While these latter techniques remain largely research-focused, they point to the expanding potential of multimodal imaging for linking retinal structure with systemic disease processes.
• Figure 3: An annotated fundus image reproduced from ref65, showing pathological markers including microaneurysm, haemorrhage, exudates, and anatomical regions like the fovea and optic disc.
• Figure 4: illustrates a timeline that outlines the development of artificial intelligence, starting with symbolic rule-based systems in the 1950s and progressing to modern deep learning, foundation models, and large language models. This evolution emphasises important technological changes that have impacted the overall AI field, providing a basis for today’s uses in healthcare and biomedical research.
• Figure 5: Before the rise of oculomics, computational pipelines for ophthalmology followed a straightforward path: Images were acquired as input for clarity, processed through pre-processing. The classical method went through feature engineering with a need for manual interpretation, including contrast enhancement, region of interest cropping and segmentation, and noise reduction. Then these were put in the model for training, testing and validating the model for the specific ocular diseases for machine learning and early deep learning modelling. This work is in the classification, detection and prediction of different diseases. These systems focused almost entirely on ocular diseases such as cataract, glaucoma, age-related macular degeneration, diabetic retinopathy, and corneal opacity. While they laid the foundation for computer-assisted diagnosis, they were limited in scope, largely handcrafted, and confined to the eye itself without considering wider systemic health.