Semi-Automated Measurement of Central Retinal Artery Blood Flow Using Doppler Ultrasound Coupled With Electrocardiography
- 1Department of Ophthalmology, Icahn School of Medicine at Mount Sinai, New York, NY, USA
- 2Maine College of Engineering and Computing, The University of Maine, Orono, ME, USA
- 3Department of Cardiology, Icahn School of Medicine at Mount Sinai, New York, NY, USA
Abstract
Purpose: To design, implement, and test a semi-automated method for measuring central retinal artery (CRA) blood flow using color Doppler imaging (CDI) waveforms synchronized with electrocardiography (ECG).
Methods: Twenty-four patients (48 eyes; 16/8 male/female; mean age 64.5 ± 15.8 years) underwent CDI-ECG imaging. Peak systolic velocity (PSV), end-diastolic velocity (EDV) and time from the ECG r-wave to the CRA PSV (r_PSV) were measured manually and with a semi-automated method.
Results: PSV showed excellent correlation and agreement between manual and semi-automated methods (right eye [OD] r = 0.95, intraclass correlation coefficient [ICC] = 0.945; left eye [OS] r = 0.94, ICC = 0.928). EDV had weaker results (OD r = 0.67, ICC = 0.441; OS r = 0.86, ICC = 0.585). The r_PSV had moderate correlations (OD r = 0.75; OS r = 0.42) and agreement (ICC = 0.702 and 0.387). Bland-Altman analysis showed small PSV bias (OD = 0.327; OS 0.419 cm/s), EDV overestimation (bias OD −0.556; OS −0.886 cm/s), and minimal r_PSV bias (OD = −0.012; OS = −0.010 s). Mixed-effects modeling found a significant method and eye effect only for EDV.
Conclusions: Semi-automated PSV measurements show excellent agreement with manual analysis and represent the most robust and clinically applicable output of the method. In contrast, EDV and r_PSV demonstrate lower agreement and should be considered exploratory, likely reflecting sensitivity to waveform morphology and low-magnitude signal characteristics.
Translational Relevance: A semi-automated CDI-ECG waveform analysis could allow for reproducible assessment of ocular blood flow and enable large-scale assessment of data examining the heart-eye relationship.
Related
- A comparative study of the state-of-the-art artificial intelligence models for automated plaque measurement in a carotid ultrasound Japanese cohort
- Automated Segmentation of Retinal Pigment Epithelium Layers in High-Resolution OCT
- Central retinal artery occlusion preceding hemispheric ischemic stroke treated with mechanical thrombectomy : A case report
- Effect of Intravitreal Anti-VEGF Injections on the Vitreomacular Interface in Treatment-Naïve Eyes With Central Retinal Vein Occlusion
- A Semi-Automated Interactive Annotation Framework for Intraoperative Neurosurgical Ultrasound: A Proof-of-Concept Study
BibTeX
@article{wood2026semiautomated,
title = {{Semi-Automated Measurement of Central Retinal Artery Blood Flow Using Doppler Ultrasound Coupled With Electrocardiography}},
author = {Keren Wood and Alon Harris and Temitope Olayinka and Alice Verticchio Vercellin and Yoav Niv-Granot and Stamatios Lerakis and Chen Gurevitz and Brent Siesky and Yash Lahoti and Giovanna Guidoboni},
journal = {Translational Vision Science \& Technology},
year = {2026},
month = oct,
doi = {10.1167/tvst.15.10.43},
url = {https://doi.org/10.1167/tvst.15.10.43},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - Semi-Automated Measurement of Central Retinal Artery Blood Flow Using Doppler Ultrasound Coupled With Electrocardiography
AU - Keren Wood
AU - Alon Harris
AU - Temitope Olayinka
AU - Alice Verticchio Vercellin
AU - Yoav Niv-Granot
AU - Stamatios Lerakis
AU - Chen Gurevitz
AU - Brent Siesky
AU - Yash Lahoti
AU - Giovanna Guidoboni
PY - 2026
DA - 2026/10/27
JO - Translational Vision Science & Technology
DO - 10.1167/tvst.15.10.43
UR - https://doi.org/10.1167/tvst.15.10.43
AB - Purpose: To design, implement, and test a semi-automated method for measuring central retinal artery (CRA) blood flow using color Doppler imaging (CDI) waveforms synchronized with electrocardiography (ECG). Methods: Twenty-four patients (48 eyes; 16/8 male/female; mean age 64.5 ± 15.8 years) underwent CDI-ECG imaging. Peak systolic velocity (PSV), end-diastolic velocity (EDV) and time from the ECG r-wave to the CRA PSV (r_PSV) were measured manually and with a semi-automated method. Results: PSV showed excellent correlation and agreement between manual and semi-automated methods (right eye [OD] r = 0.95, intraclass correlation coefficient [ICC] = 0.945; left eye [OS] r = 0.94, ICC = 0.928). EDV had weaker results (OD r = 0.67, ICC = 0.441; OS r = 0.86, ICC = 0.585). The r_PSV had moderate correlations (OD r = 0.75; OS r = 0.42) and agreement (ICC = 0.702 and 0.387). Bland-Altman analysis showed small PSV bias (OD = 0.327; OS 0.419 cm/s), EDV overestimation (bias OD −0.556; OS −0.886 cm/s), and minimal r_PSV bias (OD = −0.012; OS = −0.010 s). Mixed-effects modeling found a significant method and eye effect only for EDV. Conclusions: Semi-automated PSV measurements show excellent agreement with manual analysis and represent the most robust and clinically applicable output of the method. In contrast, EDV and r_PSV demonstrate lower agreement and should be considered exploratory, likely reflecting sensitivity to waveform morphology and low-magnitude signal characteristics. Translational Relevance: A semi-automated CDI-ECG waveform analysis could allow for reproducible assessment of ocular blood flow and enable large-scale assessment of data examining the heart-eye relationship.
SN - 2164-2591
ER -