Title: Automated Erythrocyte Detection and Tracking for Retinal Blood Flow Quantification in Erythrocyte-Mediated Angiography

Abstract:

Capillary-level retinal blood flow (RBF) holds significant promise as a diagnostic biomarker for a range of ocular conditions. Despite this potential, current technologies capable of measuring RBF at the capillary level are scarce. Erythrocyte-mediated angiography (EMA), a nascent imaging modality, facilitates such measurements by rendering individual red blood cells visible. However, the automated detection and tracking of these cells—prerequisites for accurate blood flow quantification—have received limited attention. To bridge this gap, we introduce EMTrack, an innovative framework designed for erythrocyte detection and tracking. EMTrack incorporates a flow-context module that effectively differentiates between stationary and moving cells, alongside a topology-aware tracking strategy engineered to handle significant motion variations and large displacements between frames. Furthermore, we present RBF-EMA, a novel EMA dataset enriched with detailed annotations for both erythrocyte detection and tracking. Our experimental evaluations reveal that EMTrack surpasses existing baseline methods in both quantitative metrics and qualitative performance on the RBF-EMA dataset. Additionally, the RBF quantification outcomes underscore the robust capability of our proposed framework for the automated assessment of retinal blood flow.

Source: arXiv Generated at: 2026-06-02 00:00:00 UTC