Signal Strength as an Important Factor in the Analysis of Peripapillary Microvascular Density Using Optical Coherence Tomography Angiography

The quality of the scan image is important in peripapillary circulation analysis using optical coherence tomography angiography (OCTA). We aimed to investigate the effects of signal strength (SS) on the peripapillary microvascular density acquired from OCTA. A total of 259 eyes from 259 young healthy subjects were included. Peripapillary OCTA images using 3 × 3 mm angiography scan were acquired from all participants. Subjects were divided into four groups according to the SS: SS 7, SS 8, SS 9, and SS 10. Vessel density (VD) and perfusion density (PD) of the superficial capillary plexus were calculated. VD and PD were compared among the four groups, and linear regression analyses were performed to identify and evaluate the clinical factors associated with average VD. As the SS increased from 7 to 10, the average VD and PD increased; these increases were statistically significant (all, p < 0.001). Regression analyses showed that four factors were significantly correlated with average VD: age (partial r = 0.133), average retinal nerve fiber layer thickness (partial r = 0.169), cup/disc ratio (partial r =−0.481), and SS (partial r = 0.413). SS is a significant factor affecting peripapillary microvascular density, and its influence is similar to well-known structural parameters associated with glaucoma.

peripapillary microvascular perfusion has yet to be determined. Because the vessel distributions of macular and peripapillary areas considerably differ, information on the effects of SS on OCTA measurements in the peripapillary area is required.
Therefore, we investigated the effects of SS on peripapillary microvascular density measured from OCTA to determine the appropriate level for peripapillary perfusion analyses. Results patient characteristics. In total, 150 eyes of males and 109 eyes of females were enrolled. The number of eyes in each group was 99 (SS 7), 82 (SS 8), 42 (SS 9), and 36 (SS10). The mean ages of the SS 7, SS 8, SS 9, and SS 10 groups were 26.62 ± 8.55, 26.80 ± 8.38, 25.84 ± 6.71, and 27.56 ± 5.26 years, respectively. No statistically significant differences were found in age and sex among the four groups. There were no significant differences in ocular characteristics including the BCVA and spherical equivalent, IOP, central corneal thickness, axial length, central macular thickness, and RNFL thickness. The baseline characteristics of the participants are shown in Table 1.

Discussion
The present study confirmed that peripapillary microvascular density is significantly affected by SS. This trend is similar to that observed in our previous study. In brief, we previously performed 6 × 6 mm macular angiography scans and found an association between SS and macular microvascular density 26 . VD and PD values increased with as the SS increased from 7 to 9. No difference was evident between SS 9 and SS 10 in any sector, with the exception of several of the outer sectors. Venugopal et al. 27 also reported that SS had a significant effect on the repeatability of OCTA with the VD values increasing in scans with higher SS values.
In this study, the VD and PD patterns observed for the four SS groups differed, depending on the sector. The superior, temporal, and inferior sectors in the SS 8, 9, and 10 groups showed no difference, but were higher than  those in the SS 7 group, whereas the VD and PD in the nasal sector significantly increased as the SS rose from 7 to 10. On average, significant differences were found among the SS 7, 8, and 9 groups, but no difference was observed between the SS 9 and 10 groups, which was attributed to summing over the four sectors. The results of this study differed from those of our previous study for the macular area. First, the vessel distributions in the peripapillary and macular areas were completely different. Large vessels from the optic disc, forming a retinal vessel arcade, comprised a large portion of the 3 × 3 mm angiography scan in the peripapillary area, whereas there were microvasculature and foveal avascular zones in the macular area and no large vessels. In the process of calculating the VD and PD, repeated OCT B-scans were obtained from the same location, and motion contrast was produced by calculating decorrelation/dissimilarity of repeated B-scans. To remove the invalid pixels from the OCTA image which was associated with noisy or low OCTA pixels, the thresholding algorithm was applied to the acquired unthresholded OCTA B-scans. Finally, the thresholded OCTA scan image was displayed 28 . In this process, large and small vessels may show different results, depending on the SS. Large vessels are less affected by SS, whereas signals from small vessels may be attenuated or not detected ( Fig. 2) 8 . For these reasons, the results of this study differed from those of the previous study, and different patterns were observed in the four peripapillary sectors. The second reason is the difference in scanning area. In our previous study, a 6 mm macular scan was performed. The 6 × 6 mm and 3 × 3 mm scans have different scan densities and number of repetitive scans 29 .   Table 2. Comparison of the vessel density among groups with a signal strength (SS) from 7 to 10 (mm −1 ). Mean ± SD. * The p-value was obtained from one-way ANOVA. † The p-value was obtained from the post hoc test (Bonferroni) between SS 7 and SS 8 groups. ‡ The p-value was obtained from the post hoc test (Bonferroni) between SS 8 and SS 9 groups. § The p-value was obtained from the post hoc test (Bonferroni) between SS 9 and SS 10 groups. Boldface numbers indicate statistically significant differences at p < 0.05.  www.nature.com/scientificreports www.nature.com/scientificreports/ Currently, optic disc and peripapillary perfusion OCTA analyses are used to evaluate many diseases, many of which involve glaucoma. Jia et al. 11 reported that the disc flow index was reduced by 25% in the glaucoma group, and Liu et al. 9 also reported decreased peripapillary VD by 13% in glaucomatous eyes (93.00% in normal eyes, 80.55% in glaucomatous eyes). Some investigators also reported that deep retinal layer and parapapillary choroidal microvasculature dropout is associated with RNFL thinning in primary open-angle glaucoma 13,19 . Other studies have reported a reduction in optic disc perfusion in glaucoma 10,12,[14][15][16][17][18][30][31][32] . In addition, Wang et al. 33 indicated a 12.5% reduction in the disc perfusion index of multiple sclerosis with the ON group compared to control eyes. Another study observed a significant reduction of the peripapillary VD in LHON patients (−6.2%, −10.2%, and −25.7% reduction compared to controls in the early acute stage, late-acute stage, and chronic stage, respectively) 20 . However, in most of the aforementioned studies, SS was only used as study inclusion criteria; a comparison of SS between disease and control groups was not performed. In this study, SS 7, 8, and 9 groups showed decreased values compared to the SS 10 group by 13.8%, 9.4%, and 3.8% in VD and 15.6%, 11.1%, and 4.7% in PD, respectively (Fig. 2). Although the OCTA device used in each study was different, considering the calculation process of microvascular density in OCTA as mentioned above, the effects of SS influence in the above studies cannot be excluded completely.
According to previous studies, optic disc perfusion was significantly associated with visual field defects, as well as structural measures such as the RNFL, ganglion cell-inner plexiform layer (GC-IPL), and rim area in glaucoma patients. Yarmohammadi et al. 32 reported that circumpapillary VD was significantly associated with average RNFL thickness (r = 0.73) and rim area (r = 0.44), and another study found an association with RNFL (r = 0.406) and GC-IPL (r = 0.413) 34 . This trend was also observed in other studies 11,35 . In this study, age, rim area, disc area, and SS were significantly correlated with VD values, similar to previous studies. In multivariate regression analyses, the rim area was not significantly associated with average VD; however, this may be due to the interaction between average RNFL thickness and rim area. Partial r in SS was 0.413, the second highest absolute value after cup/disc ratio (r = −0.481). Thus, SS plays a significant role and should be considered, along with the well-known structural factors associated with glaucoma.
Scan image quality is essential for the quantitative assessment of microvascular density. Various factors such as patient cooperation, media opacity, and operator skill can affect the image quality of OCT scan. OCTA images with a higher SS show better clarity and improved segmentation error, which can increases the reliability of the measurement. Additionally, previous studies have reported that quantitative metrics in the OCTA is significantly correlated with the SS 36,37 . If OCTA scan image is obtained with low SS, the repeatability and reproducibility are reduced, and the OCTA measurements become unreliable. For accurate data analyses, recommended SS values which is defined by the manufacturer's proprietary guidelines should be obtained. For example, Cirrus HD-OCT recommends ≥6 (signal strength, 0 to 10), Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany) recommends more than 15 (Q score, 0 to 40), and RTVue (Optovue, Fremont, CA, USA) recommends ≥45 (signal strength index, 0 to 100) for macular patterns and ≥35 for retinal scanning 38 . However, there has still been no discussion of the optimal SS level in OCTA analyses. All subjects enrolled in the present study were healthy, young volunteers who did not have any ocular or systemic abnormalities. In addition, the four groups were compared under the same conditions (e.g., age, sex, and various ocular factors). Therefore, the effects of SS on the peripapillary microvascular density are considered a meaningful finding, and the method of microvascular density estimation in various articles previously reported should be considered in correlation with SS.
This study had several limitations. First, the effects of SS were not verified on various OCTA devices. Although the OCTA measurement method is similar in that it uses the decorrelation of moving particles in vessels, the commercial OCTA devices currently used are based on different imaging algorithms; thus, the effects of SS on measured values may vary from one device to another. Second, the effects of SS were examined in normal healthy subjects; thus, its effects in diseased eyes are unknown. Third, the numbers of SS 9 and 10 groups were slightly lower than those of SS 7 and 8 groups. Finally, to investigate the microvasculature of the optic nerve head, it may be helpful to exclude large vessels using various methods such as manual segmentation or applying a specific www.nature.com/scientificreports www.nature.com/scientificreports/ threshold algorithm, but we did not perform that. Since it is considered that modifications of scan image would not be useful in clinical practice, we used built-in software only. Additional studies are needed to overcome these limitations.
In conclusion, we investigated the effect of SS on the OCTA measurements in the young healthy subjects and found that peripapillary VD has a significant correlation with SS. Therefore, when analyzing average VD and PD, it is necessary to try to obtain the highest SS possible. Considering our results, an SS of at least 9 in the average sector and 8 in superior, temporal, and inferior sectors may be needed in the analysis of peripapillary microvascular density. SS is a significant factor that affects peripapillary microvascular density, and its influence is similar to the well-known structural parameters of glaucoma. Our study is the first to analyze the effects of SS on peripapillary VD. The results from this study are expected to be particularly useful to evaluate peripapillary perfusion, and clinicians should consider SS when interpreting optic disc OCTA scans.

Methods
This was a prospective cohort study. The study protocol was approved by the Armed Forces Medical Command Institutional Review Board (Seongnam, Republic of Korea) and followed to the tenets of the Declaration of Helsinki. All subjects who met the eligibility criteria provided written informed consent to participate.
Subjects. This study initially included 588 eyes from 294 healthy, young volunteers who visited the Armed Forces Capital Hospital for a health screening checkup. All subjects underwent comprehensive ophthalmic examinations including a slit-lamp biomicroscopy, refraction,best-corrected visual acuity (BCVA), intraocular pressure (IOP). Additionally, central corneal thickness (NT-530, NIDEK Co., LTD, Gamagori, Aichi, Japan), axial (lower row, C,D) in the same subject. The average peripapillary vessel density (VD) and perfusion density (PD) of SS 7 were reduced as compared with SS 9 by 23.78% and 23.77%, respectively. Small vessels in nasal area was well detected in SS 9 (arrow head), whereas signals was significantly attenuated in SS 7 (arrow). length (IOLMaster, Carl Zeiss Meditec, Dublin, CA, USA) were measured, and OCT and OCTA using the Zeiss Cirrus 5000 (Carl Zeiss Meditec, Dublin, CA, USA) was performed. All subjects underwent a 3 × 3 mm angiography scan in the peripapillary area using the AngioPlex OCTA device from the Zeiss Cirrus 5000 system, without pupil dilation, and no eyedrops or gels were used to degrade the image quality. The excluded subjects were those with a history of systemic disease including diabetes and hypertension, a history of neuro-ophthalmic or retinal disease, media opacity, glaucoma, a history of ocular trauma, an IOP > 21 mmHg, BCVA < 20/25, OCTA scan with an SS < 7, a, and the presence of a segmentation error in the OCTA scan. If both eyes met the inclusion criteria, one eye was randomly selected. Finally, 259 eyes from 259 healthy, young subjects were included in this study.
OCT and OCTA. The Zeiss HD-OCT 5000 instrument with an AngioPlex OCTA device was used to acquire microvasculature images of the peripapillary area. This instrument operated at a central wavelength of 840 nm and a speed of 68,000 A-scans per second. In the 3 × 3 mm scan pattern, there were 245 A-scans in each B-scan along the horizontal dimension and 245 B-scan positions along the vertical dimension. Each B-scan was repeated four times at the same position 29 . The optical microangiography-complex algorithm analyzed the change in complex signals (both intensity and phase changes contained within sequential B-scans performed at the same position) 39,40 , and then generated en face microvascular images. The vascular images of the superficial capillary plexus (SCP), which spanned from the internal limiting membrane to the inner plexiform layer, and the deep capillary plexus, which extended from the inner nuclear layer to the outer plexiform layer, were displayed separately. The AngioPlex incorporated the FastTrac retinal tracking technology to reduce motion artifacts. All scans were analyzed using Cirrus OCTA software (AngioPlex software, version 10.0; Carl Zeiss Meditec). VD (defined as the total length of the perfused vasculature per unit area in the region of measurement) and PD (defined as the total area of the perfused vasculature per unit area in the region of measurement) of the SCP in the 3 × 3 mm scan were measured automatically 41 ; the scan consisted of a centered 1 mm diameter circle and four outer sectors including superior, nasal, inferior, and temporal areas (Fig. 3A,B). All scans were performed by the same experienced examiner and reviewed individually by two investigators (H.B.L. and Y.W.K.) for quality evaluation (i.e., SS, segmentation error, loss of fixation, and motion artifacts), and substandard scans were excluded. The central foveal thickness (CFT), retinal nerve fiber layer (RNFL) thickness, and optic disc parameters (including the rim area and cup/disc ratio) were measured using a macular cube 512 × 128 combination scan and an optic disc cube 200 × 200 scan.
Statistical analyses. Subjects were divided into four groups according to the SS of the angiography scan: SS 7, 8, 9, and 10. The continuous variables among the four groups were compared by one-way analysis of variance (ANOVA) with a post-hoc Bonferroni correction, and the categorical variables were compared using the χ2 test. The VD and PD in the peripapillary area were also compared among the four groups using one-way ANOVA with a post hoc Bonferroni correction. Univariate and multivariate linear regression analyses were performed to identify and evaluate the clinical factors associated with average peripapillary VD. All statistical analyses were performed using SPSS statistical software for Windows, version 21.0 (SPSS, IBM Corp, Armonk, NY, USA) and GraphPad Prism TM (version 6.0; GraphPad Software, San Diego, CA, USA). Snellen BCVA results were converted into the logarithm of the minimum angle of resolution value. Continuous variables are presented as the mean ± standard deviation. Differences were considered significant at p < 0.05.