Artificial intelligence-based quantification of retinal microvascular biomarkers from fundus photography of chronic kidney disease: a case-control study - Report - MDSpire
Advertisement
Artificial intelligence-based quantification of retinal microvascular biomarkers from fundus photography of chronic kidney disease: a case-control study
Clinical Report: AI-Driven Analysis of Retinal Microvascular Indicators in CKD
Overview
This study identifies retinal microvascular parameters associated with chronic kidney disease (CKD) and evaluates their discriminatory ability using AI-based fundus image analysis. Key findings indicate that specific retinal features can serve as non-invasive indicators for early CKD detection.
Background
Chronic kidney disease (CKD) is a significant global health issue, affecting over 10% of the population and often remaining undiagnosed in early stages. The lack of accessible and non-invasive risk identification tools contributes to this challenge. Retinal microvascular changes may reflect systemic alterations associated with CKD, presenting an opportunity for early detection through non-invasive methods.
Data Highlights
Parameter
Odds Ratio (OR)
P-value
Lower Arteriovenous Ratio (AVR)
0.617
0.006
Reduced Arterial Tortuosity (aTort)
0.380
<0.001
Lower Arterial Vascular Density (aVD)
0.642
0.027
Higher Venous Vascular Density (vVD)
1.910
<0.001
Higher Vessel Tortuosity (VT3PD)
2.020
0.012
Key Findings
Lower AVR, reduced aTort, and lower aVD are associated with CKD.
Higher vVD and VT3PD indicate positive associations with CKD.
The final model demonstrated moderate discrimination for CKD (AUC = 0.776).
Exploratory analysis showed limited discrimination between early and advanced CKD (AUC = 0.748).
Retinal microvascular alterations may be detectable in early-stage CKD.
Clinical Implications
The findings suggest that retinal microvascular parameters can serve as non-invasive indicators for early CKD detection, potentially improving screening practices. Clinicians may consider incorporating retinal imaging into routine assessments for at-risk populations to enhance early identification of CKD.
Conclusion
AI-derived retinal microvascular parameters show promise as non-invasive indicators of early CKD. Further validation in larger populations is necessary to confirm these findings and their clinical utility.
