Abstract
Purpose: :
To determine the extent to which suboptimal image focus might skew measurement of retinal vessel caliber. Studies have utilized central retinal arteriolar and venular calibers (CRAE & CRVE), measured from images using IVAN software (UW-Madison), to detect risk factors for systemic diseases. Rochtchina et al. (2008) found systematic differences in caliber between racial groups: in eyes with greater pigmentation, measured CRAE & CRVE are greater, even after adjustment for other factors. One hypothesis is that this is due to racial difference in RPE/blood vessel contrast and/or suboptimal focus rather than an anatomic variation. We explored differential image focus as a possible contributing factor to larger vascular caliber measurements in eyes with greater RPE pigmentation.
Methods: :
From the reading center database, 20 digital retinal images were selected for optimum quality (focus, illumination) - 10 from Caucasians and 10 from African Americans (judging from skin/iris pigmentation in anterior images). Adobe Photoshop CS3 (Box Blur) was used to derive 9 progressively more blurred versions of the originals - all still within the normal focus range observed at our reading center. IVAN (v1.3) was used to measure CRAE & CRVE in the originals and their blurred series.
Results: :
Median CRAE = 154µm and CRVE = 237µm across groups. Comparing CRAE and CRVE between blurred variants and original images with the Wilcoxon signed rank test we found significant difference (p<0.01) in caliber at all blur levels in both racial groups (except Blur 1/Caucasian, P=0.05). Measured vascular caliber increased at higher blur levels. CRAE exceeded a threshold of +5µm at Blur 3-4 and CRVE at Blur5, and approached +20µm at Blur7 for both vessel types.
Conclusions: :
Suboptimal image focus artificially broadens measured vessel caliber - regardless of racial group. Differences reported by Rochtchina were CRAE - 11µm, CRVE - 18µm. If there are systematic reasons for more suboptimal focus in some racial groups (darker/lower contrast images are harder to focus), this effect may contribute to racial differences in vessel measurement.
Keywords: clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology • imaging/image analysis: clinical • retina