October 2016
Volume 57, Issue 13
Open Access
Physiology and Pharmacology  |   October 2016
Measurements of Retinal Perfusion Using Laser Speckle Flowgraphy and Doppler Optical Coherence Tomography
Author Affiliations & Notes
  • Nikolaus Luft
    Department of Clinical Pharmacology, Medical University of Vienna, Vienna, Austria
    Department of Ophthalmology, Kepler University Hospital, Linz, Austria
    University Eye Hospital, Ludwig-Maximilians University, Munich, Germany
  • Piotr A. Wozniak
    Department of Clinical Pharmacology, Medical University of Vienna, Vienna, Austria
    Department of Ophthalmology, Medical University of Warsaw, Warsaw, Poland
  • Gerold C. Aschinger
    Center for Medical Physics and Biomedical Engineering, Medical University of Vienna, Vienna, Austria
    Institute of Applied Physics, Vienna University of Technology, Vienna, Austria
  • Klemens Fondi
    Department of Clinical Pharmacology, Medical University of Vienna, Vienna, Austria
  • Ahmed M. Bata
    Department of Clinical Pharmacology, Medical University of Vienna, Vienna, Austria
  • René M. Werkmeister
    Department of Clinical Pharmacology, Medical University of Vienna, Vienna, Austria
  • Doreen Schmidl
    Department of Clinical Pharmacology, Medical University of Vienna, Vienna, Austria
    Center for Medical Physics and Biomedical Engineering, Medical University of Vienna, Vienna, Austria
  • Katarzyna J. Witkowska
    Department of Clinical Pharmacology, Medical University of Vienna, Vienna, Austria
  • Matthias Bolz
    Department of Ophthalmology, Kepler University Hospital, Linz, Austria
  • Gerhard Garhöfer
    Department of Clinical Pharmacology, Medical University of Vienna, Vienna, Austria
  • Leopold Schmetterer
    Department of Clinical Pharmacology, Medical University of Vienna, Vienna, Austria
    Center for Medical Physics and Biomedical Engineering, Medical University of Vienna, Vienna, Austria
    Singapore Eye Research Institute, the Academia, Singapore
    Lee Kong Chian School of Medicine, Nanyang Technological University, Novena Campus, Singapore
  • Correspondence: Leopold Schmetterer, Department of Clinical Pharmacology, Medical University of Vienna, Währinger Gürtel 18-20, 1090 Vienna, Austria; leopold.schmetterer@meduniwien.ac.at
Investigative Ophthalmology & Visual Science October 2016, Vol.57, 5417-5425. doi:10.1167/iovs.16-19896
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Nikolaus Luft, Piotr A. Wozniak, Gerold C. Aschinger, Klemens Fondi, Ahmed M. Bata, René M. Werkmeister, Doreen Schmidl, Katarzyna J. Witkowska, Matthias Bolz, Gerhard Garhöfer, Leopold Schmetterer; Measurements of Retinal Perfusion Using Laser Speckle Flowgraphy and Doppler Optical Coherence Tomography. Invest. Ophthalmol. Vis. Sci. 2016;57(13):5417-5425. doi: 10.1167/iovs.16-19896.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose: This study evaluated the validity of retinal perfusion measurements using laser speckle flowgraphy (LSFG) by means of in vitro experiments and direct comparison with dual-beam Doppler optical coherence tomography (D-OCT) in a healthy Caucasian population.

Methods: The flow velocity of scattering solution pumped through a glass capillary was measured at 17 different flow velocities (range, 0.5–47 mm/s) using LSFG. The flow within the glass capillary was produced by a computer-controlled infusion pump. In vivo, three consecutive LSFG scans were obtained in 20 eyes of 20 healthy Caucasian subjects before and after pharmacological pupil dilation. Relative flow volume (RFV), the primary output parameter of LSFG, was comparatively validated relative to absolute measurements of retinal blood flow and velocity as obtained from D-OCT.

Results: In the in vitro experiments, RFV was found to saturate at a level of approximately 700 arbitrary units (au) or 23.5 mm/s of actual velocity. In vivo, RFV was in significant agreement with absolute blood flow measurements as obtained from D-OCT in arteries (r = 0.69, P = 0.001) and veins (r = 0.74, P < 0.001). However, linear regression analysis revealed significant positive zero offset values for RFV of 223.4 and 282.7 au in arteries and veins, respectively.

Conclusions: Measurements of RFV were successfully obtainable, reproducible, and not influenced by pharmacological pupil dilation. Nevertheless, our data revealed flaws in the LSFG method of measuring retinal perfusion in Caucasians. Adjustment to the technique is required to address apparent issues with RFV, especially saturation effects with higher arterial flow rates. The present dataset may provide a valuable tool to do so. (Clinicaltrials.gov number NCT02582411).

Retinal and choroidal perfusion abnormalities have been reported to play an important role in the pathogenesis of various ocular diseases including glaucoma, age-related macular degeneration, diabetic retinopathy, and central serous chorioretinopathy.13 Hence, it is urgent that a valid, reliable and clinically feasible technique for the measurement of retinal and choroidal blood flow in humans be developed. The introduction of bidirectional laser Doppler velocimetry (LDV)4,5 made it possible to derive absolute quantification of retinal blood flow in vivo in the 1970s and 1980s, but the technique is time consuming and requires considerable operator experience. Doppler Fourier-domain optical coherence tomography (D-OCT)69 has overcome some of the technical limitations inherent in LDV and enables valid and robust measurements of absolute blood flow in selected retinal vessels9,10 and in total retinal blood flow.7,11 Although qualitative visualization of retinal perfusion has been realized in a commercially available OCT system (Envisu C-Class unit; Leica Microsystems, Wetzlar, Germany), nevertheless, no commercial OCT-based technique is currently available for the quantitative measurement of retinal perfusion. 
In recent years, laser speckle flowgraphy (LSFG) has emerged as a promising method for the measurement of ocular perfusion in humans1217 and was recently granted clearance by the US Food and Drug Administration for quantitative evaluation of blood flow in retinal vessels.18 LSFG permits noninvasive measurements of blood flow at the optic nerve head (ONH), the retina, and the choroid, simultaneously, with a patient-friendly acquisition time of less than 5 s. LSFG-derived readings of retinal perfusion, however, are determined not only by the blood flow velocity in the retinal vasculature but also by the perfusion of the underlying choroidal tissue due to the long wavelength (830 nm) of the applied laser source.19 Recently, relative flow volume (RFV) was introduced as a novel LSFG-derived blood flow index in an effort to exclude the choroidal background noise from the measurements of perfusion in retinal vessels.15 The first comparison with LDV indicated favorable correlation with absolute blood flow in the human retina.15 To date, however, the in vivo experience with RFV is restricted to Japanese subjects. 
In the present study, we aimed to study the validity of RFV as a measurement of retinal blood flow by means of in vitro experiments as well as direct comparison with D-OCT in a healthy Caucasian population. The presented work is the first to apply LSFG to measure retinal circulation in Caucasians and to validate this method by comparison with D-OCT-derived measurements of retinal circulation. 
Methods
Laser Speckle Flowgraphy
In the present study, a commercially available LSFG system (LSFG-NAVI; Softcare Co., Ltd., Fukuoka, Japan) was used to quantify perfusion in a custom-built flow phantom and in retinal vessels in vivo. The principles of LSFG have previously been outlined in detail.2024 The LSFG device used in this study consisted of a fundus camera supplied with an 830-nm diode laser and a digital charge-coupled device camera (750 × 360 pixels). The fundamental output parameter of LSFG is mean blur rate (MBR). This parameter constitutes a measurement of relative blood flow velocity and is expressed in arbitrary units (au). During one LSFG scan, a total of 118 images are continuously captured at a rate of 30 frames per second over a time period of approximately 4 s. The device is equipped with dedicated analysis software (LSFG Analyzer version 3.1.58; Softcare), which synchronizes and mathematically averages all 118 acquired images to produce a so-called “composite map” depicting the distribution of perfusion in the ocular fundus within one cardiac cycle. In this color-coded map, a rectangular band is centered on the retinal vessel of interest, using the built-in image analysis software (Fig. 1). Within the selected area, the retinal vessel is automatically delineated by computing a threshold between MBR values in the retinal vessel and the background MBR originating from perfusion of the underlying choroid.15 The vessel diameter determined by LSFG is expressed in pixels. Relative flow volume (in arbitrary units) represents the primary output parameter of LSFG for the analysis of retinal perfusion. This index of blood flow in retinal vessels is automatically calculated by subtracting the background MBR in the nonvessel area (i.e., choroidal MBR) from the MBR detected within the retinal vessel area.15 
Figure 1
 
Color-coded “composite” map depicts the distribution of blood flow in the ocular fundus (left). Rectangular bands (RBs) were centered on the retinal vessels of interest in vicinity to the optic nerve head margin (white dots) using the equipped image analysis software. The MBR waveform (right) shows the synchronized and averaged MBR signal within one cardiac cycle. The green curve represents the MBR signal obtained within RB 1 (artery), and the purple curve represents the MBR signal of RB 2 (vein). MBR, mean blur rate; RB, rectangular bands.
Figure 1
 
Color-coded “composite” map depicts the distribution of blood flow in the ocular fundus (left). Rectangular bands (RBs) were centered on the retinal vessels of interest in vicinity to the optic nerve head margin (white dots) using the equipped image analysis software. The MBR waveform (right) shows the synchronized and averaged MBR signal within one cardiac cycle. The green curve represents the MBR signal obtained within RB 1 (artery), and the purple curve represents the MBR signal of RB 2 (vein). MBR, mean blur rate; RB, rectangular bands.
In Vitro Capillary Measurements
In order to test the validity of quantitative perfusion estimation with LSFG, a model eye in conjunction with a flow phantom was deployed. Thereby, a 4.0-mm artificial pupil was aligned with a simple lens with a focal length of 30.0 mm and a glass capillary with an inner diameter of 300 μm placed in the focal plane of this model eye. In order to simulate blood flow within the capillary, a constant and continuous flow of diluted milk was produced by an infusion pump (Pilot C model; Fresenius, Homburg, Germany). This computer-controlled system provides a high-flow-rate accuracy of ±2% (as stated by the manufacturer). A wide range of 17 different flow velocities (range, 0.5–47 mm/s) was produced, and LSFG was applied to obtain readings of RFV three consecutive times at each velocity. The mean value of three successive measurements was used for comparison with preset velocities. In an in vivo setting, RFV is intended to account for the underlying choroidal perfusion that codetermines the MBR obtained within the area of retinal vessels. Even though choroidal background perfusion was not incorporated into our flow phantom, we used RFV for direct comparison with preset velocities in order to account for the offset (i.e., the MBR value measured outside of the area of the glass capillary, where there is no blood flow).15,19,21,25 
In Vivo Measurements on Retinal Vessels
Subjects.
This prospective comparison study included 20 healthy Caucasian subjects recruited by the Department of Clinical Pharmacology of the Medical University of Vienna. The study protocol was approved by the Ethics Committee of our institution. After providing subjects an explanation of the nature and possible consequences of the study, all subjects gave written informed consent. All research adhered to guidelines outlined in the Declaration of Helsinki. 
All subjects underwent an extensive screening examination during the two weeks prior to the actual study day that comprised medical history, physical examination, best-corrected visual acuity testing using standard Early Treatment of Diabetic Retinopathy Study charts, slit-lamp examination including indirect fundoscopy, measurement of IOP, using Goldmann applanation tonometry, measurements of systolic blood pressure (SBP) and diastolic blood pressure (DBP) with automated oscillometry and a urine pregnancy test for women. Any of the following criteria excluded a subject from participation in the study: smoking, ametropia ≥ 6 diopters (D), contact lens wearing, any ocular surgery within 3 previous months, significant opacities of the optical medium (e.g., corneal scars, lens opacities classification system II (LOCS-II) grading ≥ 3, vitreous opacities), any other relevant ocular disease or abnormality, any clinically relevant illness as judged by the investigators, uncontrolled hypertension with systolic blood pressure (SBP) ≥160 mm Hg and/or diastolic blood pressure (DBP) ≥ 100 mm Hg, current or planned pregnancy or lactation and blood donation in the previous 3 weeks. If both eyes of a subject were eligible, one eye was randomly selected to be the study eye. All subjects abstained from alcohol and stimulating beverages containing xanthine derivatives (tea, coffee, cola-like drinks) for at least 12 hours before measurements were taken. Mean arterial blood pressure and ocular perfusion pressure measured in a seated position were calculated using appropriate formulae as follows: [Mean arterial pressure = SBP + 1/3 (SBP – DBP), ocular perfusion pressure = 2/3 MAP – IOP]. 
Protocol.
On the study day, all measurements were scheduled to be made between 7:30 AM and 5.00 PM. Before the LSFG-based measurements of retinal perfusion were performed, all subjects rested in a quiet, air-conditioned room for 10 minutes, with the temperature maintained at 22°C to 23°C. First, three successive LSFG scans were acquired while the eye was in a state of miosis. Thereafter, the pupil of the study eye was dilated by topical instillation of 0.5% tropicamide eye drops (Mydriaticum Agepha Augentropfen; Agepha Ges.m.b.H., Vienna, Austria). After a further resting period of 20 minutes, a second set of 3 consecutive LSFG scans was performed. All scans were acquired by one of three experienced examiners (NL, PAW, KF). Before each LSFG scan, the subjects were instructed to blink and then to hold their eyes open and remain still during the scan acquisition. Between the three consecutive scans, the subjects were asked to lean back, and the chinrest was readjusted. 
Subsequently, three consecutive D-OCT scans were performed in the eye in a state of mydriasis by one experienced examiner (GCA), blinded to the subjects' identity and to the result of the LSFG examination. In order to ensure LSFG and D-OCT measurements were obtained at identical retinal locations, an LSFG-derived fundus image was used to guide the D-OCT measurements. To obtain a scan, the subject was seated comfortably in front of the D-OCT system, and the head was stabilized using an adapted slit lamp headrest. As with LSFG, subjects were encouraged to blink shortly before the scanning and to keep their eyes open during signal acquisition. All LSFG and D-OCT examinations were acquired in near darkness under identical ambient conditions. Finally, measurements of IOP and systemic hemodynamics (SBP and DBP) were taken. 
Laser Speckle Flowgraphy Measurements.
In each subject, one major retinal artery and accompanying vein was measured simultaneously within the same LSFG scanning field at a distance of <1 ONH diameter from the ONH margin (Fig. 1). The mean values for vessel diameter and RVF of the three successive LSFG measurements performed in mydriasis were used for statistical analysis. 
Doppler Optical Coherence Tomography Measurements.
A dual-beam bidirectional D-OCT system was used in the present study, as described in detail previously.7,9,10 The device consists of a broadband superluminescent diode with a central wavelength of 840 nm (spectral bandwidth, 54 nm) and two charge-coupled device cameras with a maximum readout rate of 20 kHz. The system provides a resolution (in tissue) of approximately 6 and 18 μm in axial and transversal directions, respectively. The retinal vessel under study is illuminated by two probe beams under a known angle Δα that is separated by their polarization properties. Light backscattered and back-reflected from the sample (i.e., the retinal vessel under study) is spectrally detected by two identical spectrometers. Signal post processing, that is, calculation of the phase shift due to moving scatterers within the sample (i.e., erythrocytes moving in retinal vessels), was carried out in semiautomated manner, using custom software (written in LabVIEW; National Instruments; Austin, TX, USA) as outlined previously.9 
Of the three consecutive D-OCT measurements, the highest quality scan, as determined by the examiners, was selected for signal postprocessing. D-OCT gives readings of blood flow (microliters per minute) as well as blood flow velocity (millimeters per second). Furthermore, measurements of vessel caliber (in micrometers) were obtained from the OCT phase images as described in detail elsewhere.26 With respect to vessel caliber measurements using this method, Fondi et al.26 recently demonstrated high interobserver reproducibility and a high level of agreement with vessel caliber measurements as obtained from fundus images. Signal post processing was performed independently by two experienced examiners (NL and PAW), both of whom were both blinded to the subject's identity and the result of the preceding LSFG scans. For blood flow, blood flow velocity, and vessel caliber, the mean values of the two signal postprocessing procedures were used for statistical analysis. 
Statistical Analysis
Descriptive data are mean ± SD. A P value of <0.05 was considered statistically significance. All statistical analysis was performed using commercially available software (version 22.0.0.1, SPSS Statistical Software [IBM, Armonk, NY, USA] for Macintosh [Hilliard, OH, USA]). 
As a first step, histogram frequency analysis and the Shapiro-Wilk test were carried out to confirm the normality of data distribution. With respect to the in vitro experiments, the relationship between preset flow velocity and RFV was evaluated using linear regression analysis. The chi square test was used to elucidate the effect of pupil dilation on the success rate of LSFG scans in vivo. Furthermore, linear regression analysis was applied to assess the relationships between LSFG- and D-OCT-derived blood flow indices as well as vessel caliber. The ratios of the vessel calibers measured with D-OCT to vessel calibers determined by LSFG and the average measured flow were compared between arteries and veins by using appropriate Student t-tests. Measurement repeatability in vitro and in vivo was assessed by calculating the coefficient of variation (COV [%]) defined as the mean SD of repeated measurements-to-the overall mean ratio. As a second index of measurement repeatability, the intraclass correlation coefficient (ICC) for the three consecutive measurements was determined using a repeated measures ANOVA model. Finally, paired sample Student t-tests were used to compare RFV before and after pupil dilation. 
Results
In Vitro
The relationship between preset velocity rates of diluted milk in the glass capillary and RFV as obtained with LSFG is shown in Figure 2A. RFV was found to saturate at approximately 700 au, corresponding to 23.5 mm/s of actual velocity. Below this saturation level, a significant linear correlation between RFV and velocity was observed (r = 0.92, P < 0.001) (Fig. 2B). With respect to the repeatability of RFV measurements in the flow phantom, we found a COV of 8.2% and an ICC value of 0.992. 
Figure 2
 
(A) Relationship between the preset velocity of scattering solution with RFV as obtained with LSFG in vitro. The dotted red lines indicate the presumed saturation level of RFV at approximately 700 au and 23.5 mm/s, respectively. (B) Linear regression analysis between preset velocity and RFV for the range of 0 to 700 au and 0 to 23.5 mm/s, respectively.
Figure 2
 
(A) Relationship between the preset velocity of scattering solution with RFV as obtained with LSFG in vitro. The dotted red lines indicate the presumed saturation level of RFV at approximately 700 au and 23.5 mm/s, respectively. (B) Linear regression analysis between preset velocity and RFV for the range of 0 to 700 au and 0 to 23.5 mm/s, respectively.
In Vivo
Subjects.
A total of 10 male and 10 female Caucasian subjects were included. A summary of participants' baseline characteristics is provided in Table 1. All subjects showed clear optical medium with no opacities in the crystalline lens (LOCS-II grade, 0). Histogram frequency analysis and Shapiro-Wilk test results confirmed the normal distribution of the analyzed parameters (data not shown). 
Table 1
 
Demographic and Baseline Characteristics of Subjects
Table 1
 
Demographic and Baseline Characteristics of Subjects
LSFG Measurement Success Rate.
Without pupil dilation, neither of the three successive LSFG scans of suitable retinal artery and accompanying vein were obtainable in one subject due to insufficient pupil size. After pharmacological pupil dilation, 4 scans of 3 subjects were excluded due to fixation compliance issues. Hence, the success rate of RFV measurements was 95% (57 of 60) with natural (scotopic) pupils and 93.3% (56 of 60) after pharmacological pupil dilation (P = 0.70). 
Correlation of LSFG with D-OCT.
Linear regression analysis showed that RFV was significantly correlated with absolute blood flow as obtained with D-OCT in arteries (r = 0.69, P = 0.001) (Fig. 3A) and in veins (r = 0.74, P < 0.001) (Fig. 3B). Furthermore, RFV demonstrated a significant correlation with blood flow velocity as determined with D-OCT in veins (r = 0.52, P = 0.02) (Fig. 3D). In contrast, this association did not reach the level of statistical significance in arteries (r = 0.38, P = 0.10) (Fig. 3C). The respective regression equations revealed considerable positive zero offsets for RFV ranging between 227.4 and 282.7 (Figs. 3A–3D). Mean RFV as determined by LSFG was higher in veins (461.7 ± 106.8) than in arteries (348.8 ± 72.0), with a resulting 32.4% relative difference (P < 0.001). Mean flow as determined by D-OCT was 8.1 ± 3.2 μl/min in arteries and 8.6 ± 3.8 μl/min in veins with a (nonsignificant) difference of 5.6% (P = 0.69). 
Figure 3
 
Correlation of absolute blood flow as obtained from D-OCT with the laser speckle flowgraphy-derived parameter RFV in arteries (A) and veins (B). Correlation between blood flow velocity as determined by D-OCT and RFV is shone in arteries (C) and veins (D).
Figure 3
 
Correlation of absolute blood flow as obtained from D-OCT with the laser speckle flowgraphy-derived parameter RFV in arteries (A) and veins (B). Correlation between blood flow velocity as determined by D-OCT and RFV is shone in arteries (C) and veins (D).
Vessel Calibers.
Vessel calibers as obtained from D-OCT phase images were significantly correlated with the vessel calibers as determined by LSFG in arteries (r = 0.87, P < 0.001) (Fig. 4A) and veins (r = 0.81, P < 0.001) (Fig. 4B). However, the ratio of the vessel calibers determined by D-OCT (in micrometers) to those measured using LSFG (in arbitrary units) was significantly higher for retinal arteries (8.98 ± 0.74 μm/au) than for veins (8.38 ± 0.74 μm/au), with a P value <0.01. This finding suggests a relative underestimation of the vessel lumen by LSFG in arteries compared with that in veins. 
Figure 4
 
Correlation between vessel caliber as determined by Doppler optical coherence tomography (micrometers) and by laser speckle flowgraphy (pixels) in retinal arteries (A) and veins (B).
Figure 4
 
Correlation between vessel caliber as determined by Doppler optical coherence tomography (micrometers) and by laser speckle flowgraphy (pixels) in retinal arteries (A) and veins (B).
As shown in Figure 5, RFV was significantly correlated with vessel caliber as determined by LSFG in arteries (r = 0.59, P = 0.006) (Fig. 5A) and veins (r = 0.84, P < 0.001) (Fig. 5B). Figure 6 presents the correlations of D-OCT-derived blood flow with the vessel caliber as obtained from the OCT phase images. Larger vessel calibers were significantly associated with higher levels of blood flow, both in arteries (r = 0.71, P < 0.001) (Fig. 6A) and in veins (r = 0.91, P < 0.001) (Fig. 6A). 
Figure 5
 
Relationship between vessel caliber and relative flow volume as determined by laser speckle flowgraphy in arteries (A) and veins (B), respectively.
Figure 5
 
Relationship between vessel caliber and relative flow volume as determined by laser speckle flowgraphy in arteries (A) and veins (B), respectively.
Figure 6
 
Correlation between blood flow and vessel caliber as determined by Doppler optical coherence tomography in arteries (A) and veins (B), respectively.
Figure 6
 
Correlation between blood flow and vessel caliber as determined by Doppler optical coherence tomography in arteries (A) and veins (B), respectively.
Repeatability.
Table 2 summarizes the calculated repeatability indices COV and ICC for both the RFV and the vessel diameter in arteries and veins, obtained before and after pharmacological pupil dilation, respectively. Similar indices of repeatability with COVs of 5.59% or lower and ICCs of 0.973 or higher were observed independently of pupil dilation. Hence, pharmacological pupil dilation did not affect the repeatability of LSFG-based measurements of RFV and vessel diameter. 
Table 2
 
Repeatability Indices for Relative Flow Volume and Vessel Diameter as Determined by LSFG
Table 2
 
Repeatability Indices for Relative Flow Volume and Vessel Diameter as Determined by LSFG
Effect of Pupil Dilation.
Before pupil dilation, the mean RFV was 345 ± 74 au in the selected retinal arteries and 469 ± 108 au in veins. After pharmacological pupil dilation, mean RFVs of 345 ± 72 au and 466 ± 108 au were obtained in arteries and veins, respectively. Paired-sample Student t-tests revealed no significant effect of pupil dilation on the measurements of RFV in arteries (P = 0.90) or veins (P = 0.69). 
Discussion
The primary purpose of this study was to evaluate the validity of LSFG as a method to quantify retinal perfusion. For this purpose, as a first step, an in vitro experiment was conducted wherein we performed LSFG measurements of a scattering solution pumped through a glass capillary. We found a saturation effect for RFV at approximately 700 au or 23.5 mm/s of actual velocity. Even though regression analysis indicated reasonable linear correlation of RFV with preset velocity below this saturation level (r = 0.92), in fact, the relationship clearly appeared nonlinear. 
Second, in an in vivo experiment, we performed measurements of perfusion in major retinal vessels, using LSFG and D-OCT. Noninvasive quantification of retinal blood flow in vivo is challenging, and currently, no gold standard method exists. Nevertheless, the validity of D-OCT to measure absolute retinal blood flow has been largely substantiated using a number of in vitro and in vivo approaches.9,10,2730 It is worth highlighting a recent study by Told et al.,11 which found very good concordance between retinal blood flow measurements obtained with D-OCT and with the terminal microsphere method in a nonhuman primate model. The latter method represents an independent invasive technique and is considered the gold standard of blood flow validation.31 
The main output parameter of LSFG when analyzing retinal perfusion, RFV, serves as a nondimensional index of blood flow in the retinal vessel examined. To our knowledge, only Shiga et al.15 previously endeavored to validate RFV by using another measuring technique. They applied LDV in 30 retinal arteries and detected good correlations between RFV and blood flow velocity (r = 0.61) and absolute blood flow (r = 0.51), as determined from LDV, respectively. Moreover, in their study, the sum of RFV values in the two daughter vessels was strongly correlated with RFV in the trunk vein (r = 0.98). However, their work failed to acknowledge an apparent problem with measurements of RFV. In their study, the regression equation exhibited a considerable negative zero offset for RFV, indicating that RFV readings were zero when blood flow obtained with LDV was still 4.5 μl/min, which constituted more than 40% of their mean arterial flow (11.2 ± 3.8 μl/min). The zero offset in the regression analysis between RFV and LDV-based velocity readings was similar, with 17.3 mm/s or approximately 48% of the average arterial velocity of 36.1 ± 7.9 mm/s. 
In the present report, we demonstrated that RFV significantly correlated with absolute blood flow as obtained from D-OCT, both in retinal arteries (r = 0.69) and veins (r = 0.74). Nevertheless, the relationship between RFV and blood flow velocity was only significant in veins (r = 0.52), not in arteries (r = 0.38). Surprisingly, the respective regression equations indicated that RFV would hypothetically amount to 223.4 au in arteries and 282.7 au in veins if the D-OCT flow reading was zero. Similar results were observed for blood flow velocity. This is in opposition to the data of Shiga et al.,15 who found negative zero offsets for RVF compared to LDV.15 The underlying reasons for this discrepancy could not be fully elucidated in the present study. Nevertheless, we presume that the positive zero offset of RFV in this study was mainly due to an underestimation of blood flow by LSFG with higher flow rates. This study provides multiple pieces of evidence in support of this hypothesis. 
First, our in vitro experiments clearly revealed a saturation effect for RFV at approximately 23.5 mm/s of actual velocity, a level that was well within the range of velocity observed in vivo (range, 5.3–28.1 mm/s). Rose and Hudson32 previously demonstrated comparable levels of blood velocity in a healthy population (32.04 ± 2.14 and 23.97 ± 1.68 mm/s in retinal arterioles and venules, respectively) with arterial peak systolic velocities reaching 55.56 ± 3.61 mm/s. Even though the data from our in vitro experiment cannot be directly extrapolated to an in vivo situation, this finding substantiates the fact that there exists an upper limit to blood flow velocity that can be measured with this technique and that this limit lies within the physiologic range of retinal blood flow velocity cannot be excluded. To our knowledge, no study has so far addressed the maximum measurable blood flow velocity for LSFG. The laser speckle method essentially constitutes a “decorrelation” technique, which uses the phenomenon that the speckle pattern reflected from moving scatterers (e.g., cellular blood components) “de-correlates” with a static speckle pattern. In theory, once the maximum decorrelation is reached, a further increase in actual blood flow velocity will not result in higher RFV readings. As a result, the mismatch between the actual blood flow velocity and the detected variation in the laser speckle pattern will naturally increase with higher flow rates. 
A further potential technical limitation of the LSFG method lies in its two-dimensional measurement principle. Because LSFG does not permit depth-resolved measurements of perfusion, the relative attribution of the different depth layers within the vessel to the obtained absolute RFV signal is unclear. In contrast, D-OCT allows for precise depth gating of flow information. Hence, it may be that, especially in vessels with larger calibers and higher velocities, the deeper layers of blood flow do not fully contribute to the RFV signal, resulting in a relative underestimation of flow by LSFG compared with D-OCT. Murray's law predicts that blood flow increases with vessel caliber raised to the power of three (D3) in vascular systems with an optimum compromise between blood flow and vascular resistance.33 Thus, it can be deduced that there should exist a linear correlation between retinal flow velocity and vessel caliber in subjects with normal retinal perfusion. This has been established by recent reports from our group, using LDV4 and D-OCT9 in healthy subjects as well as by previous laser Doppler velocimetry data.34 In keeping with these previous data, the present study showed that blood flow as determined by D-OCT strongly depends on vessel caliber in arteries (r = 0.71) and veins (r = 0.91). The respective correlations between RFV and vessel caliber were weaker with r = 0.59 in arteries and r = 0.84 in veins. 
Second, our data indicate that LSFG may have an inherent problem with respect to measurement of vessel caliber in retinal arterioles. High-blood-flow rates promote axial migration of erythrocytes to the center of flow and an accumulation of erythrocytes at the center of the vessel, hence, causing a cell-free layer of blood plasma close to the wall of the vessel lumen.35 Accordingly, it was previously demonstrated that retinal arteries exhibit a nonuniform distribution of MBR with higher values observed at the center of the vessel and lower values close to the vessel wall, which may lead to an underestimation of the vessel lumen.25 Even though the vessel calibers as determined by LSFG and D-OCT were in close correlation in the present study, we detected a relative underestimation of the vessel lumen by LSFG in arteries in relation to veins. This is in agreement with a recent report by Iwase et al.,25 who made identical findings from comparisons of lumen diameters obtained by LSFG and those obtained by an adaptive optics camera. 
In the third place, LSFG readings of venous flow were significantly higher than those of RFV in paired arteries by 32.4%. This is in contrast to our D-OCT data obtained for the identical vessels as well as to findings from previous investigations that applied LDV, which clearly demonstrated comparable levels of blood flow between retinal arteries and veins.4,5,36 Interestingly, our findings are in concordance with those of a recent experiment by Iwase et al.25 In their LSFG-based study, total retinal flow index was calculated as the sum of RFVs from all major retinal vessels surrounding the ONH, and a similar mismatch (25%) between total retinal flow indexes in arteries (1455 ± 348 au) and veins (1812 ± 445 au) was observed. All of these findings indicate a potential underestimation of higher (arterial) flow with LSFG. 
A further point worth considering is that the study by Shiga et al.15 included exclusively Japanese subjects, who exhibit a higher level of fundus pigmentation than Caucasians. As an index of retinal perfusion, RFV is intended to adjust for the underlying background signal originating from choroidal perfusion. RFV is calculated by subtracting the background LSFG signal in the nonvessel area (presumably corresponding to choroidal perfusion) from the signal detected within a retinal vessel. The intensity of the background signal in LSFG is to a large degree determined by the absorption of the emitted light (830 nm) by retinal pigment. Thus, it cannot be excluded that this correction method is inadequate in Caucasians, who exhibit lower levels of fundus pigmentation and, hence, higher levels of choroidal background signal in LSFG. 
Herein, we evaluated the clinical usability of a commercial LSFG device to quantitatively estimate retinal perfusion in a Caucasian population. First, we showed that a satisfactory success rate of RFV measurements (>93%) could be achieved regardless of pharmacological pupil dilation. Moreover, we detected satisfactory repeatability for readings of RFV in retinal arteries and veins with COVs of 5.59% or lower and ICCs of 0.97 or higher. In addition, readings of RFV were obtainable with similarly high repeatability independently of pupil dilation. Values for ICC can range between 0 and 1, the latter value representing perfect repeatability. An ICC value above 0.9 indicates acceptable clinical repeatability.37,38 This is in accordance with the study by Shiga et al.,15 which recently reported a COV of 5.9% for RFV obtained for temporal retinal arteries in 18 Japanese subjects. To our knowledge, only one additional report by Iwase et al.25 previously investigated the relatively novel parameter RFV in a sample of Japanese subjects. In both of those studies, measurements were performed with the eyes in a state of mydriasis. Our report was the first to demonstrate that neither the readings of RFV nor their repeatability were affected by pharmacological pupil dilation. 
Limitations to this study may be found. First, the study was limited by its small sample size and by the nature of its cross-sectional design. Furthermore, with D-OCT, the mean values of two signal postprocessing procedures by two examiners were used for statistical analysis, whereas, with LSFG, the mean value of three consecutive LSFG measurements was used. However, excellent reliability of the D-OCT system used in the present work was shown in previous reports of our group.7,9,10 The use of diluted milk rather than whole heparinized blood as the scattering solution may be regarded as a limitation to our in vitro experiment. Nevertheless, a previous in vitro experiment conducted by our group9 with a virtually identical setup demonstrated excellent agreement between D-OCT based measurements of absolute flow with the preset velocities. 
In conclusion, this comparative study of D-OCT and LSFG indicates considerable issues with LSFG-based measurements of retinal perfusion in healthy Caucasian subjects. Even though the patient friendliness and real-time nature of measurements appear to render LSFG a promising technique for the exploration of retinal perfusion status in a clinical setting, nevertheless, our in vivo and in vitro data congruently pointed toward concern for saturation effects for RFV at physiological blood flow rates. Moreover, in contrast to absolute D-OCT measurements, RFV readings were significantly different between paired arteries and veins and at low flow rates correction of values with regard to underlying choroidal signals needs to be investigated. Hence, adjustment to the technique and further validation of RFV seems mandatory. The present data obtained in Caucasian subjects may serve as a good basis to work on adaptation of the algorithms to obtain absolute flow values in the future. 
Acknowledgments
Supported by an unrestricted grant from NIDEK Co. Ltd., Japan. 
Disclosure: N. Luft, None; P.A. Wozniak, None; G.C. Aschinger, None; K. Fondi, None; A.M. Bata, None; R.M. Werkmeister, None; D. Schmidl, None; K.J. Witkowska, None; M. Bolz, None; G. Garhöfer, None; L. Schmetterer None 
References
Cherecheanu AP, Garhofer G, Schmidl D, Werkmeister R, Schmetterer L. Ocular perfusion pressure and ocular blood flow in glaucoma. Curr Opin Pharmacol. 2013; 13: 36–42.
Pemp B, Schmetterer L. Ocular blood flow in diabetes and age-related macular degeneration. Can J Ophthalmol. 2008; 43: 295–301.
Saito M, Saito W, Hirooka K, et al. Pulse waveform changes in macular choroidal hemodynamics with regression of acute central serous chorioretinopathy. Invest Ophthalmol Vis Sci. 2015; 56: 6515–6522.
Garhofer G, Werkmeister R, Dragostinoff N, Schmetterer L. Retinal blood flow in healthy young subjects. Invest Ophthalmol Vis Sci. 2012; 53: 698–703.
Riva CE, Grunwald JE, Sinclair SH, Petrig BL. Blood velocity and volumetric flow rate in human retinal vessels. Invest Ophthalmol Vis Sci. 1985; 26: 1124–1132.
Baumann B, Potsaid B, Kraus MF, et al. Total retinal blood flow measurement with ultrahigh speed swept source/Fourier domain OCT. Biomed Opt Express. 2011; 2: 1539–1552.
Doblhoff-Dier V, Schmetterer L, Vilser W, et al. Measurement of the total retinal blood flow using dual beam Fourier-domain Doppler optical coherence tomography with orthogonal detection planes. Biomed Opt Express. 2014; 5: 630–642.
Wang Y, Bower BA, Izatt JA, Tan O, Huang D. In vivo total retinal blood flow measurement by Fourier domain Doppler optical coherence tomography. J Biomed Opt. 2007; 12: 041215.
Werkmeister RM, Dragostinoff N, Palkovits S, et al. Measurement of absolute blood flow velocity and blood flow in the human retina by dual-beam bidirectional Doppler fourier-domain optical coherence tomography. Invest Ophthalmol Vis Sci. 2012; 53: 6062–6071.
Werkmeister RM, Dragostinoff N, Pircher M, et al. Bidirectional Doppler Fourier-domain optical coherence tomography for measurement of absolute flow velocities in human retinal vessels. Opt Lett. 2008; 33: 2967–2969.
Told R, Wang L, Cull G, et al. Total retinal blood flow in a nonhuman primate optic nerve transection model using dual-beam bidirectional Doppler FD-OCT and microsphere method. Invest Ophthalmol Vis Sci. 2016; 57: 1432–1440.
Aizawa N, Yokoyama Y, Chiba N, et al. Reproducibility of retinal circulation measurements obtained using laser speckle flowgraphy-NAVI in patients with glaucoma. Clin Ophthalmol. 2011; 5: 1171–1176.
Cull G, Told R, Burgoyne CF, Thompson S, Fortune B, Wang L. Compromised optic nerve blood flow and autoregulation secondary to neural degeneration. Invest Ophthalmol Vis Sci. 2015; 56: 7286–7292.
Kiyota N, Shiga Y, Takahashi H, Nakazawa T. Large vessel area of the optic nerve head, measured with laser speckle flowgraphy is significantly reduced in eyes with preperimetric glaucoma. Clin Exp Ophthalmol. 2015; 43: 841–843.
Shiga Y, Asano T, Kunikata H, et al. Relative flow volume a novel blood flow index in the human retina derived from laser speckle flowgraphy. Invest Ophthalmol Vis Sci. 2014; 55: 3899–904.
Shiga Y, Omodaka K, Kunikata H, et al. Waveform analysis of ocular blood flow and the early detection of normal tension glaucoma. Invest Ophthalmol Vis Sci. 2013; 54: 7699–7706.
Wang L, Burgoyne CF, Cull G, Thompson S, Fortune B. Static blood flow autoregulation in the optic nerve head in normal and experimental glaucoma. Invest Ophthalmol Vis Sci. 2014; 55: 873–880.
U.S. Food and Drug Administration. Section 510(k) Premarket Notification Number K153239. May 12, 2016. Available at: http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm?ID=K153239. Accessed July 4 2016.
Nagahara M, Tamaki Y, Tomidokoro A, Araie M. In vivo measurement of blood velocity in human major retinal vessels using the laser speckle method. Invest Ophthalmol Vis Sci. 2011; 52: 87–92.
Briers JD, Fercher AF. Retinal blood-flow visualization by means of laser speckle photography. Invest Ophthalmol Vis Sci. 1982; 22: 255–259.
Sugiyama T, Araie M, Riva CE, Schmetterer L, Orgul S. Use of laser speckle flowgraphy in ocular blood flow research. Acta Ophthalmol. 2010; 88: 723–729.
Sugiyama T, Utsumi T, Azuma I, Fujii H. Measurement of optic nerve head circulation: comparison of laser speckle and hydrogen clearance methods. Jpn J Ophthalmol. 1996; 40: 339–343.
Tamaki Y, Araie M, Kawamoto E, Eguchi S, Fujii H. Noncontact two-dimensional measurement of retinal microcirculation using laser speckle phenomenon. Invest Ophthalmol Vis Sci. 1994; 35: 3825–2834.
Tamaki Y, Araie M, Tomita K, Nagahara M, Tomidokoro A, Fujii H. Real-time measurement of human optic nerve head and choroid circulation using the laser speckle phenomenon. Jpn J Ophthalmol. 1997; 41: 49–54.
Iwase T, Ra E, Yamamoto K, Kaneko H, Ito Y, Terasaki H. Differences of retinal blood flow between arteries and veins determined by laser speckle flowgraphy in healthy subjects. Medicine (Baltimore). 2015; 94: e1256.
Fondi K, Aschinger GC, Bata AM, et al. Measurement of retinal vascular caliber from optical coherence tomography phase images. Invest Ophthalmol Vis Sci. 2016; 57: OCT121–OCT129.
Blatter C, Coquoz S, Grajciar B, et al. Dove prism based rotating dual beam bidirectional Doppler OCT. Biomed Opt Express. 2013; 4: 1188–1203.
Tan O, Liu G, Liang L, et al. En face Doppler total retinal blood flow measurement with 70 kHz spectral optical coherence tomography. J Biomed Opt. 2015; 20: 066004.
Trasischker W, Werkmeister RM, Zotter S, et al. In vitro and in vivo three-dimensional velocity vector measurement by three-beam spectral-domain Doppler optical coherence tomography. J Biomed Opt. 2013; 18: 116010.
Werkmeister RM, Palkovits S, Told R, et al. Response of retinal blood flow to systemic hyperoxia as measured with dual-beam bidirectional Doppler Fourier-domain optical coherence tomography. PLoS One. 2012; 7: e45876.
Nilsson SFE, Alm A. Determination of ocular blood flows with the microsphere method. In: Schmetterer L, Kiel J. eds. Ocular Blood Flow. Berlin: Springer-Verlag; 2012: 25–47.
Rose PA, Hudson C. Comparison of retinal arteriolar and venular variability in healthy subjects. Microvasc Res. 2007; 73: 35–38.
Murray CD. The Physiological principle of minimum work. I. The vascular system and the cost of blood volume. Proc Natl Acad Sci U S A. 1926; 12: 207–214.
Riva C, Ross B, Benedek GB. Laser Doppler measurements of blood flow in capillary tubes and retinal arteries. Invest Ophthalmol. 1972; 11: 936–944.
Tateishi N, Suzuki Y, Tanaka J, Maeda N. Imaging of oxygen saturation and distribution of erythrocytes in microvessels. Microcirculation. 1997; 4: 403–412.
Feke GT, Tagawa H, Deupree DM, Goger DG, Sebag J, Weiter JJ. Blood flow in the normal human retina. Invest Ophthalmol Vis Sci. 1989; 30: 58–65.
Kramer MS, Feinstein AR. Clinical biostatistics. LIV. The biostatistics of concordance. Clin Pharmacol Ther. 1981; 29: 111–123.
Muller R, Buttner P. A critical discussion of intraclass correlation coefficients. Stat Med. 1994; 13: 2465–2476.
Figure 1
 
Color-coded “composite” map depicts the distribution of blood flow in the ocular fundus (left). Rectangular bands (RBs) were centered on the retinal vessels of interest in vicinity to the optic nerve head margin (white dots) using the equipped image analysis software. The MBR waveform (right) shows the synchronized and averaged MBR signal within one cardiac cycle. The green curve represents the MBR signal obtained within RB 1 (artery), and the purple curve represents the MBR signal of RB 2 (vein). MBR, mean blur rate; RB, rectangular bands.
Figure 1
 
Color-coded “composite” map depicts the distribution of blood flow in the ocular fundus (left). Rectangular bands (RBs) were centered on the retinal vessels of interest in vicinity to the optic nerve head margin (white dots) using the equipped image analysis software. The MBR waveform (right) shows the synchronized and averaged MBR signal within one cardiac cycle. The green curve represents the MBR signal obtained within RB 1 (artery), and the purple curve represents the MBR signal of RB 2 (vein). MBR, mean blur rate; RB, rectangular bands.
Figure 2
 
(A) Relationship between the preset velocity of scattering solution with RFV as obtained with LSFG in vitro. The dotted red lines indicate the presumed saturation level of RFV at approximately 700 au and 23.5 mm/s, respectively. (B) Linear regression analysis between preset velocity and RFV for the range of 0 to 700 au and 0 to 23.5 mm/s, respectively.
Figure 2
 
(A) Relationship between the preset velocity of scattering solution with RFV as obtained with LSFG in vitro. The dotted red lines indicate the presumed saturation level of RFV at approximately 700 au and 23.5 mm/s, respectively. (B) Linear regression analysis between preset velocity and RFV for the range of 0 to 700 au and 0 to 23.5 mm/s, respectively.
Figure 3
 
Correlation of absolute blood flow as obtained from D-OCT with the laser speckle flowgraphy-derived parameter RFV in arteries (A) and veins (B). Correlation between blood flow velocity as determined by D-OCT and RFV is shone in arteries (C) and veins (D).
Figure 3
 
Correlation of absolute blood flow as obtained from D-OCT with the laser speckle flowgraphy-derived parameter RFV in arteries (A) and veins (B). Correlation between blood flow velocity as determined by D-OCT and RFV is shone in arteries (C) and veins (D).
Figure 4
 
Correlation between vessel caliber as determined by Doppler optical coherence tomography (micrometers) and by laser speckle flowgraphy (pixels) in retinal arteries (A) and veins (B).
Figure 4
 
Correlation between vessel caliber as determined by Doppler optical coherence tomography (micrometers) and by laser speckle flowgraphy (pixels) in retinal arteries (A) and veins (B).
Figure 5
 
Relationship between vessel caliber and relative flow volume as determined by laser speckle flowgraphy in arteries (A) and veins (B), respectively.
Figure 5
 
Relationship between vessel caliber and relative flow volume as determined by laser speckle flowgraphy in arteries (A) and veins (B), respectively.
Figure 6
 
Correlation between blood flow and vessel caliber as determined by Doppler optical coherence tomography in arteries (A) and veins (B), respectively.
Figure 6
 
Correlation between blood flow and vessel caliber as determined by Doppler optical coherence tomography in arteries (A) and veins (B), respectively.
Table 1
 
Demographic and Baseline Characteristics of Subjects
Table 1
 
Demographic and Baseline Characteristics of Subjects
Table 2
 
Repeatability Indices for Relative Flow Volume and Vessel Diameter as Determined by LSFG
Table 2
 
Repeatability Indices for Relative Flow Volume and Vessel Diameter as Determined by LSFG
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×