Abstract
Purpose :
To evaluate parafoveal microvascular anatomy using fractal dimension (FD) analysis in normal healthy volunteers using two optical coherence tomography angiography (OCTA) machines. FD is a geometrical index of complexity in self-similar structures, which can be applied to the complexity of retinal vascular microanatomy scanned on OCTA.
Methods :
Parafoveal 3mm x 3mm OCTA scans were performed in 50 eyes of 25 healthy volunteers using two commercially available machines, Optovue RTVue XR Avanti (OA) (Optovue, Fremont CA, USA) and Zeiss Cirrus HD-OCT 5000 (ZC) (Carl Zeiss Meditec, Inc., Dublin CA, USA). Superficial and deep plexuses were identified by automated segmentation, and skeletonized images of OA and ZC were obtained using imageJ (National Institutes of Health, Bethesda MD, USA) (Figure 1) and employing identical skeletonization technique for scans from both OA and ZC. FD analysis for all images was achieved on ImageJ using FracLac plugin. Eyes of same patient were analyzed in paired fashion using student t-test, and interclass correlation coefficient (ICC) was calculated between machines. Statistical analyses were performed with SPSS, Version 23 (SPSS, Inc, Chicago, Illinois, USA)
Results :
Mean subject age was 33 years and there were 14 (56%) females. By OCT, mean central macular thickness was significantly greater on OA vs. ZC (259 vs. 257μm, p=0.023). By OCTA, both the superficial and deep plexus FD values were significantly different (OA vs. ZC) (1.816±0.0022 vs. 1.800±0.0044 for superficial, p<0.0001) (1.815±0.0034 vs. 1.791± 0.0054 for deep, p<0.0001), respectively. Poor correlation between machines was evident in both superficial (ICC=0.083) and deep (ICC=0.072) plexuses. Table 1.
Conclusions :
FD analysis of OCTA scans is a novel method for assessment of retinal microvascular anatomy. Eyes imaged on two commercial OCT angiography platforms showed significant differences using FD method which could be related to the concentration of pixels per image and the number of B-scans per machine.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.