Purchase this article with an account.
Miriam Englander, Eleftherios Paschalis, Tamara Lee, Amit Vasanji, Rishi Singh; Retinal Vascular Measurements Using Automated Fundus Image Analysis: The Effect of Hypertension on Vascular Caliber in Adult Population. Invest. Ophthalmol. Vis. Sci. 2013;54(15):23.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To implement an automated fundus image analysis in order to determine the effects of hypertension (HTN) on the measurement of retinal arteriole and venule caliber.
This is a retrospective study of 1,481 adults in an executive health screening program. All clinical data was collected from 2007-2009. Nonmydriatic retinal images were obtained during screening examinations using the Topcon NW6s digital camera. Automated measurements of vascular caliber were generated from retinal images using a validated algorithm. Retinal arteriole caliber and venule caliber were compared between normotensive and hypertensive patients and correlated to the systolic and diastolic blood pressure.
The 1,481 patients enrolled were divided to two groups: normotensive (82.3%) and hypertensive (17.7%). The mean age was 53 ± 9.5 years (mean ± SD) 80% were male, 79% were Caucasian and 21% were African Americans, Hispanic, Latin, and Asian. Both groups were similar in demographics, while only the hypertensive group was treated with antihypertensive medications. There was significant difference in the mean systolic and diastolic blood pressure between the groups; in the normotensive group (n=1,218) the mean pressures were 123mmHg ± 13.3 and 75mmHg ± 8.7 respectively, and in the hypertensive group (n=263) the mean pressures were 135mmHg ± 15.6 and 81mmHg ± 9.6 respectively (p<0.001). Furthermore, group differences were also found in the arterial caliber and venule caliber (p<0.0001), and arterial-venous ratio (AVR) (p= 0.04). In the normotensive group, systolic blood pressure (SBP) was negatively correlated to the arterial and venule caliber (r= -0.192, p<0.0001; r= -0.102; p<0.001, respectively). However, this correlation was not present in the hypertensive group (p>0.4). In confirmation with previous studies, there was a significant difference between the right and left eyes with regards to total arteriole area (p<0.0001), total venule area (p<0.0001), and AVR (p<0.0001), independent of diagnosis of HTN.
This study demonstrates the successful implementation of automated measurements of metrics, such as arteriole and venule caliber, in detecting hypertensive changes in the retina vasculature. These results suggest that automated measurements of nonmydriatic retinal images may be used to supplement the diagnosis of systemic hypertension.
This PDF is available to Subscribers Only