# Three-dimensional spatiotemporal analysis for the assessment of retinal capillary perfusion using a clinical OCT system

## TL;DR

This paper presents a new method to analyze retinal capillary perfusion using OCTA data, offering more accurate insights into early signs of vision-threatening diseases.

## Contribution

The study introduces a novel protocol using unprocessed OCTA data and depth-resolved CoV analysis for retinal perfusion assessment.

## Key findings

- The proposed method provides more accurate measurements of retinal perfusion heterogeneity compared to conventional CoV analysis.
- Using unprocessed OCTA data enhances the reliability of perfusion analysis by incorporating depth-dependent signals.
- The developed techniques can support future research on retinal vascular disorders.

## Abstract

Growing evidence suggests that subtle changes in retinal microcirculation may precede structural damage in vision-threatening diseases. Among these, perfusion heterogeneity within the retinal capillary network has emerged as a promising biomarker for early detection and disease monitoring. Recent advances in optical coherence tomography (OCT) and OCT-based angiography (OCTA) have enabled high-resolution, three-dimensional imaging of retinal morphology and vasculature. However, commercial systems remain limited in their ability to accurately analyze retinal perfusion dynamics due to reliance on proprietary and undisclosed post-processing algorithms. This paper introduces an effective protocol for spatial and temporal analysis of capillary perfusion heterogeneity using unprocessed OCTA volume data acquired by a commercial retinal imaging system. The proposed method employs a novel analysis utilizing the depth-resolved pixel-wise coefficient of variation (CoV) to quantitatively estimate retinal capillary perfusion heterogeneity. Comparison between the proposed method and conventional CoV analysis emphasizes the reliability of the new approach, incorporating depth-dependent signals. By using unprocessed OCTA data, the proposed method can provide more accurate measurements of retinal perfusion heterogeneity. Furthermore, the image processing techniques developed in this study could serve as a foundation for future research in other retinal vascular disorders.