In Vivo Measurement of Blood Velocity in Human Major Retinal Vessels Using the Laser Speckle Method

Abstract

Purpose.: To develop a simple, noninvasive method of determining blood velocity and flow through human retinal vessels (RVs), by using the laser speckle method and validating the results by bidirectional laser Doppler velocimetry (LDV).

Method.: The square blur rate (SBR), a quantitative blurring index of the laser speckle pattern that parallels the velocity of moving substances, obtained from blood flowing through glass capillary tubes (RV analogues), correlated with tube diameter, background reflectance and absorption, flow velocity, and the SBR obtained from blood flowing through underlying glass capillary tubes (choroidal vessel analogues). A nomogram was constructed to calculate the blood velocity in human RVs from the SBR values obtained in vivo. Blood velocities in RVs were determined in 12 normal eyes by using the laser speckle method and bidirectional LDV. Measurements were performed twice at the same site at 1-hour intervals.

Results.: Measurements from a temporal superior artery (n = 12; mean ± SD) were blood velocity (V _mean), 41.7 ± 4.2 mm/s; flow, 13.0 ± 3.2 μL/min; and diameter, 119.5 ±15.7 μm and time to complete one measurement, 65 ± 18 seconds, with the laser speckle apparatus; and V _mean, 37.7 ± 6.7 mm/s; flow, 11.7 ±3.0 μL/min; diameter, 111.1 ±16.6 μm; and measurement time, 112 ± 25 seconds, with the bidirectional LDV apparatus. The results obtained by the two methods correlated with each other (V _mean, r = 0.59, P = 0.023; flow, r = 0.83, P = 0.005; and diameter, r = 0.56, P = 0.032). The coefficients of reproducibility for V _mean, blood flow, and diameter measurement were 9.5% ± 2.5%, 10.5% ± 3.2%, and 5.3% ± 2.7% for the former and 15.3% ± 4.2%, 18.5% ± 4.1%, and 6.2% ± 2.2% for the latter, respectively.

Conclusions.: The laser speckle method accurately and reproducibly determines blood velocity in human RVs in less time than the LDV apparatus requires.

Information regarding blood velocity and flow through retinal vessels (RVs) in eyes is highly important for understanding the physiologic and pathologic features of the retina. Noninvasive measurement of blood flow in human RVs is clinically performed by several methods. The dye dilution technique ^{1 –9} involves the administration of exogenous substances, making it difficult to perform repeat measurements at short intervals. In blue-field stimulation, ^{10 –13} measurements are taken only in the macular area of the retina and depend on subjective responses. Doppler flowmetry, which incorporates confocal scanning laser ophthalmoscopy, ^{14 –17} allows for easy estimation of retinal capillary circulation, but interindividual comparisons are difficult, because the results are in arbitrary units. Bidirectional laser Doppler velocimetry (LDV) ^{18 –24} is probably the only method that allows for noninvasive measurement of both blood velocity and flow in absolute units in major RV. The current LDV instruments, however, require fine alignment of the laser beam, maintenance of strict eye fixation, and a relatively long time for measurements, thus limiting its application to cooperative subjects. Although a tracking system was recently introduced for the apparatus, ²⁴ in our experience, these problems were not completely resolved. 

We constructed an apparatus for noninvasive and two-dimensional relative estimation of the tissue circulation using the laser speckle phenomenon in ocular tissues, including the retina, choroid, optic nerve head, and iris. ^{25 –27} The results obtained by this method correlate well with those obtained with the microsphere or hydrogas clearance methods. ^{25 –31}  

The purpose of the present study was to evaluate whether the laser speckle method also allows noninvasive estimation of the absolute blood velocity and flow in a human major RVs, similar to the commercially available bidirectional LDV instrument equipped with a tracking system, ²⁴ and to determine which method is easier to use in clinical practice. 

Materials and Methods

Laser Speckle Methods

Laser speckle is an interference phenomenon observed when laser (coherent light) is scattered from a diffusing surface; a speckle pattern is a random pattern with properties that can only be described statistically. One of the most useful statistics is the standard deviation of the intensity distribution of the speckle pattern (δ), which is equal to the mean intensity (<I>) in an ideal condition, but less than <I> in a nonideal condition. When the laser exposure time is constant, the speckle contrast reciprocal, δ/<I>, is expressed as a function of the exposure time and the correlation time (τ_c, equation 1) and indicates the velocity of moving substances^32,33 (blood velocity when the laser is projected onto living tissue). Blood velocity is calculated as follows:  

 

where τ_c is the correlation time of temporal fluctuation in intensity, and c is the constant given by the laser exposure time.³⁴ The blood velocity, V, is determined with the relation  

 

 

The apparatus used was a modification of the one reported previously ³⁵ and incorporated the same hardware and diode laser (wavelength, 808 nm). The square blur rate (SBR[x, y]), a quantitative index of the blurring of a speckle pattern, was calculated for each pixel as follows:  

 

where x and y represent the pixel location, k represents the image sensor scan number, I(x,y,k) represents the pulse height of the output signal obtained at the horizontal xth and vertical yth pixels (x,y; x = 1, 2, 3, …, 100; y = 1, 2, 3, .., 100) in the kth scan (k = 1, 2, 3, …, 64 in 0.125 seconds), and I _mean is the mean pulse height at the horizontal xth and vertical yth pixels in the 64 scans. 

The standard deviation of I(x,y,k), SD(x,y), was calculated as follows:  

 

 

The SBR value, an index of <I>²/δ² at each pixel was calculated as follows:  

 

where C is a calibration constant obtained in an experiment using model capillary tubes through which blood flowed at known velocities, and offset is the measured SBR value obtained when there was no blood flow. 

The SBR was calculated for each pixel and divided into 50 color-coded levels, which were displayed as color graphics on a monitor showing the two-dimensional variation of the SBR level over the field of interest. The average SBR level (SBR_av) in any rectangular field of interest on a displayed color map was calculated, and the change in SBR_av over a maximum of 5 seconds was monitored for each measurement. 

All SBR data were digitally recorded on magneto-optical (MO) disks for later analysis. 

Bidirectional LDV

A blood flow velocimeter (Laser Blood Flowmeter, CLBF 100; Canon, Tokyo, Japan), equipped with a tracking system, was used to measure blood velocity and flow through the major RVs in human eyes. The velocity measurement was based on the bidirectional Doppler velocimetry method, and the CLBF details are reported elsewhere. ²⁴ The CLBF also contains a system that measures vessel diameter. ²⁴  

In Vitro Experiments

Since previous studies have indicated that speckle signals obtained from RVs are affected, not only by velocity of red cells through vessels, but also by the vessel diameter itself and the blood flow in the underlying choroid, ^36,37 the effects of these factors were evaluated in a simple optical model in which an aspheric 60-D lens (60 DCC; Nikon, Tokyo, Japan) was placed 16 mm in front of surface glass capillary tubes (GCs) of various internal diameters (Fig. 1). A blood sample was taken from one of the authors (MN), a 39-year-old healthy man, and mixed with heparin (5 IU/mL) to prevent clotting. The number of red blood cells and hematocrit of the blood sample were 454 × 10⁴ /mL and 45%, respectively, from which blood with hematocrit of 31%, 37%, or 48% was made by dividing the sample into plasma and blood cells by centrifugation. In all experiments, it was confirmed that hemolysis did not occur in the sampled blood. The mean velocity of blood flow through the GCs was regulated with a peristaltic infusion pump (PST-100R; Iwaki, Tokyo, Japan). Measurements were performed for two successive seconds and repeated three times to obtain an average value. 

Figure 1.

 

A simple model of the human eye. Glass capillary tubes (A) as CV analogues; glass capillary tube (B) as RV analogue; and film (C) as retina and retinal pigment epithelium analogue.

After confirming that the effect of absorbance of the underlying film on the SBR obtained from surface GC with internal diameter of 150 μm (RV analogue) is linear and shows some 15% difference between 5% and 30% absorbance which were reported values for retina and choroid to a laser with an 800-nm wavelength ³⁸ in white and black subjects, respectively, absorbance of the film (neutral-density filter; Fuji Film, Tokyo, Japan) was arbitrarily fixed to 20% of that in persons with Japanese background and its reflectance to 8%, according to the reported absorbance of human retinal tissue for a laser with a 711-nm wavelength. ³⁹  

Three GCs with an internal diameter of 300 μm as an analogue of choroidal vessels (CVs) were placed parallel to each other and at a right angle to the surface GC behind the film (Fig. 1). The SBR was measured in five surface GCs (RV analogues) with internal diameters of 50, 75, 100, 150, and 200 μm, and blood velocity varied between 0 and 100 mm/s, whereas blood velocity through three posteriorly placed GCs (CV analogues; Fig. 1) was set at 10 mm/s. The SBR on the center line of the surface GCs was measured with a laser power of 2 mW. Next, the blood velocity through the posteriorly placed three GCs (Fig. 1) varied between 0 and 20 mm/s. The SBR on the center line of the posterior placed GCs, 30 μm from the outer wall of a surface GC (RV analogue) with various internal diameters, and that on the center line of the surface one, where both surface and posteriorly placed GCs crossed, were measured with a laser power of 2 mW. 

Human Experiments

The study protocol was approved by the Ethics Committee of The University of Tokyo Graduate School of Medicine and adhered to the tenets of the Declaration of Helsinki (1964, 2000). Written, informed consent was obtained from all subjects after a full explanation of the nature of the study and the requirements for participation. Twelve volunteers (age range, 20–34 years) who had neither systemic nor ocular disease and had only mild refractive errors between −1 and −4 D of spherical equivalent participated in the present study. The subjects were asked to abstain from drinking coffee and alcohol and smoking for 24 hours before the examination. Each subject rested for 10 minutes in a quiet room before the examination began. The refractive error and corneal curvature were measured with an autorefractometer (Canon). The axial length of each eye was measured with an A-mode ultrasound system (AL-1000; Tomei, Tokyo, Japan). Intraocular pressure was measured by applanation tonometry (Haag Streit, Bern, Switzerland). The systolic and diastolic brachial arterial pressure and heart rate were measured with an automatic blood pressure meter (ES-51H; Terumo, Tokyo, Japan). The pupil was dilated with 1 drop of 0.4% tropicamide (Mydrin M; Santen Pharmaceuticals, Osaka, Japan). 

Measurement of Major Retinal Arterial Blood Velocity and Flow

First, the CLBF was used to measure blood velocity and flow through a major temporal retinal artery, by using bidirectional LDV. The measurement procedures were the same as previously reported. ²⁴ All LDV measurements were performed by an experienced investigator (AT), and the time to complete the measurement was recorded. A fundus photograph was taken to record the site measured. The blood flow was calculated as ½(V _mean × area), where V _mean is the average of the center line blood speed during the cardiac cycle, and the area is the cross-sectional area of the retinal artery at the measurement site. ²⁴  

After the subject rested for 10 minutes, the blood velocity and flow at the same site of the same retinal artery as that measured by LDV were measured by the laser speckle method. All laser speckle method measurements were performed with a laser power of 2 mW by an experienced investigator (YT), who did not participate in the LDV measurements, and the time needed to complete the measurement was recorded. The laser speckle measurement data were digitally stored on MO disks. Data analysis was later performed on the recorded images on MO disks, after the measurement site was confirmed on the fundus photograph. The image speckles from a pixel located on the center line at the same site of the same major temporal retinal artery as measured by LDV (Fig. 2, C) were determined, to obtain SBR_artery, and those from a pixel located approximately 30 μm from the retinal arterial edge on both sides (Fig. 2, B1 and B2) were determined by using a ruler on the display and averaged, to obtain SBR_background values. SBR_artery consisted of the SBR attributable to the major retinal artery, retinal circulation in the deeper layer, and choroidal circulation. SBR_background consisted of the SBR attributable to retinal circulation in the deeper layer and choroidal circulation. SBR_artery and SBR_background were averaged over two pulses when eye fixation was satisfactory and were defined as the mean SBR_artery and mean SBR_background, respectively. 

Figure 2.

 

Representative result of SBR measurement in a retinal artery. The image speckles from a pixel located on the center line of a major retinal artery (C) and those from a pixel located approximately 30 μm from the vascular edge on both sides (B1, B2) were recorded and averaged to obtain SBR_artery and SBR_background, respectively. V indicates an accompanying vein. ECG voltage indicates electrocardiography during examination.

Vessel diameter was measured from a digital fundus photograph according to the computer-assisted method of Suzuki, ⁴⁰ briefly summarized here. Retinal photographs were taken with a 20° angle, using a 570-nm band-pass (563–570 nm) filter and a high-resolution digital camera (1400 × 1400 pixels). One pixel corresponds to approximately 2 μm in the fundus. A mouse-controlled cursor was used to mark a line perpendicular to the center line of the vessel being measured. Three parallel lines were drawn automatically in the fundus by the computer at 100-μm intervals, and the average along the three parallel lines was calculated from which the vessel diameter in vivo was calculated based on the axial length, corneal curvature, and refraction of the subject eye. 

The blood velocity in the retinal arteries was calculated from the mean SBR_artery, mean SBR_background, and vessel diameter, assuming that relationships between the SBR from an RV analogue GC and CV analogue GCs and the diameter of an RV analogue GC obtained in the simple model eye approximate those between the SBR _artery, SBR _background, and RV diameter in the current subjects. Although the measurement target was a major temporal retinal artery in the present experiment, an image of an accompanying vein was also obtained in the same laser speckle image (Fig. 2, V). Thus, the blood flow parameters of the accompanying vein were determined with the same procedure. The blood flow was calculated as ½(V _mean× area), where V _mean is the average of the blood velocity thus calculated during the cardiac cycle, and the area is the cross-sectional area of the retinal artery or vein at the measurement site. ²⁴ The data analysis was performed by an investigator (MN) who was blind to the experimental conditions. 

LDV and laser speckle measurements were repeated at 1-hour intervals on the same day, and the reproducibility of blood velocity (flow) and diameter measurements by the two methods at the same site of the same retinal artery was calculated from these two measurement results. The reproducibility coefficient was determined as follows:  

 

where V ₁ and V ₂ represent blood velocity, flow, or diameter in the retinal artery obtained during the first and second measurements, respectively. 

Statistics

Data are reported as the mean ± SD. The Pearson correlation was used to study the relationship between the results obtained with the laser speckle method and those from bidirectional LDV. Intergroup differences were tested by paired t-test. 

Results

In Vitro Experiments

When the mean blood velocity in the GC was 0 mm/s, the SBR was not equal to 0, probably due to random movement of blood cells in the GC, and was subtracted from the SBR value measured in the following experiments. The SBR from the surface GC increased by 12% with each 1-mW increment in laser power within the range of 1.5 and 4 mW, but was little affected by variations in hematocrit between 31% and 48%, when other conditions were kept unchanged. 

The SBR obtained from the surface GC as an RV analogue at the intersection point with the posteriorly placed GCs as the CV analogue was linearly affected by (1) blood velocity through the surface GC, (2) blood velocity through the posteriorly placed GCs, and (3) the internal diameter of the former. That is, for a given internal diameter and blood velocity of an RV analogue, the relationship was linear between the SBR obtained from the RV analogue at the intersection and the blood velocity through the CV analogue. 

Figure 3A shows an example of the SBR obtained from an RV analogue with internal diameter of 150 μm with four different blood velocities (range, 8–20 mm/s), where the blood velocity through the CV analogue was set to four different speeds between 5 and 12 mm/s, and Figure 3B shows an example of the SBR obtained from RV analogues with various internal diameters, at a fixed blood velocity of 20 mm/s when the blood velocity through the CV analogue was set to five different speeds (range, 0–20 mm/s). Thus, the blood velocity through the RV analogue, its internal diameter, and the blood velocity through the CV analogue were further changed step by step, and a nomogram that illustrates the relationship between the SBR obtained from an RV analogue, its internal diameter, actual blood velocities through it, and the SBR obtained from the CV analogue in this simple model was constructed. 

Figure 3.

 

(A) An example of the SBR obtained from GC tubes with an internal diameter of 150 μm as an RV analogue in which the blood velocity was set at four speeds between 8 and 20 mm/s. The blood velocity through the three posteriorly placed CV analogues was set at four different speeds, within the range of 5 to 12 mm/s. The SBR obtained from the RV analogue at the intersection was also affected by the blood velocity through the CV analogue. Each point indicates the average of three experiments. (B) An example of the SBR obtained from five glass capillary tubes with various internal diameters as RV analogues in which the blood velocity was set at 20 mm/s when the blood velocity through the posteriorly placed CV analogue was set at five different speeds ranging from 0 to 20 mm/s. The SBR obtained from an RV analogue at the intersection correlated to both blood flow through the CV analogue and the internal diameter of the RV analogue. Each point indicates the average of three experiments.

Measurement of Blood Velocity, Flow, and Diameter of a Human Major Retinal Artery in Human Experiments

The V _mean, blood flow, and diameter of the temporal superior artery was 37.7 ± 6.7 mm/s, 11.7 ± 3.0 μL/min, and 111.1 ± 16.6 μm by bidirectional LDV. The V _mean and blood flow estimated from SBR_artery, SBR_background, and the vessel diameter, assuming the same relationship among them as in the simple model eye, averaged, respectively, 41.7 ± 4.2 mm/s, 13.0 ± 3.2 μL/min, and 119.5 ± 15.7 μm, the diameter measured by the current method (mean ± SD, n = 12). Intergroup differences between the values obtained by both methods were not significant (paired t-test, P > 0.1), whereas the intergroup correlation was significant (V _mean, r = 0.59, P = 0.023; flow, r = 0.83, P = 0.005; diameter, r = 0.56, P = 0.032), with a higher correlation for blood flow. The reproducibility coefficient for the V _mean, blood flow, and diameter measurements averaged 15.3% ± 4.2%, 18.5% ± 4.1%, and 5.3% ± 2.7%, respectively, for bidirectional LDV and 9.5% ± 2.5%, 10.5% ± 3.2%, and 6.2% ± 2.2%, respectively (mean ± SD, n = 12), for the laser speckle method. Time needed to complete one measurement was 112 ± 25 (LDV) and 64 ± 15 (laser speckle method) seconds. As far as the current instruments used and the investigator who performed measurements are concerned, the reproducibility coefficients for V _mean and blood flow were better (paired t-test, P < 0.01) for the laser speckle method and the time needed for measurement was significantly shorter (paired t-test, P < 0.01) for the laser speckle method. In the laser speckle method, the V _mean, blood flow, and diameter of an accompanying major retinal vein could be also determined using the same laser speckle image picture. The V _mean, blood flow, and diameter of the accompanying major retinal vein was 13.7 ± 3.1 mm/s, 11.1 ± 2.2 μL/min, and 131 ± 24.1 μm, respectively. The blood flow through a retinal artery and accompanying vein measured with the laser speckle method and that in the retinal artery measured with LDV in each eye are summarized in Table 1. 

Table 1.

 

Blood Flow Data by Subject and Method

Subject	Laser Speckle Method		Canon LDV
	BF in a Retinal Artery	BF in an Accompanying Vein	BF in a Retinal Artery
1	13.1	12.3	11.8
2	11.9	10.2	10.6
3	9.9	9.0	8.9
4	11.1	10.4	9.8
5	7.3	6.9	6.2
6	17.0	13.4	15.3
7	13.9	11.2	12.8
8	9.5	8.3	8.6
9	13.2	11.8	11.9
10	16.4	13.4	15.3
11	14.6	12.3	13.2
12	18.1	14.3	16.2
Mean	13.0	11.1	11.7
SD	3.2	2.2	3.0

 Data are expressed as blood flow (BF) in micro liters/minute.

Discussion

The mean blood velocity through a major RV in normal, young Japanese eyes was estimated by using the laser speckle method, assuming that nomograms constructed based on the data obtained in a simple model eye also hold in the current subjects. The blood velocity and blood flow through major retinal arteries with a mean diameter of 119.5 μm (41.7 ± 4.2 mm/s and 13.0 ± 3.2 μL/min), respectively, were in good agreement with that determined at the same site of the same eye by bidirectional LDV. The measurement of subject eyes by the laser speckle method required approximately 50% less time than LDV and had slightly better reproducibility. It must be noted, however, that blood velocity was later determined by an independent investigator based on the laser speckle image data stored on an MO disc, and the total time needed to determine blood velocity may be shorter for the LDV measurement. On the other hand, the laser speckle method allowed us to estimate blood velocity and flow, not only of the major retinal artery, but also its accompanying retinal vein at the same time, without extending the measurement time, because the laser speckle image of the accompanying major retinal vein could be obtained in the same field. 

The ratio of I _mean (x,y) to the difference between the I _mean (x,y) and the speckle intensity for successive scans of the image speckles at each pixel I (x,y,k) have been defined as the NB (normalized blur) value. ^25,26 In our former laser speckle measurement apparatus, ^25,26 we used the NB value instead of the SBR as an approximation of [I _mean(x, y)]²/[SD(x, y)]², because high-performance arithmetic and control units were not readily available at the time and it was thus difficult to calculate SBR for each pixel every 0.125-second NB value, an approximation of SBR value, had a good linear correlation with blood velocity through a glass capillary tube and the ground glass rotation speed (velocity of diffusing substances), when blood velocity or velocity of diffusing substances was not high, ^25,36 and it correlated well with the tissue blood flow rate in the retina, choroid, and iris, determined using the microsphere method or that in the optic nerve head determined by using the hydrogen gas clearance method in rabbit eyes. ^{25 –31} The NB value and the tissue blood velocity determined by the invasive methods described herein are thought to agree because the tissue blood velocities in these tissues were not so high. In the present study, we sought to estimate the blood velocity and flow through a major retinal artery where the blood velocity was relatively high, which necessitated calculations of the SBR value using high-performance arithmetic and control units incorporated into the apparatus. 

Since the effective penetration depth of the laser in the sampled blood is thought to increase as laser power increases, which would result in an increased number of moving blood cells contributing to the speckle signal, power was fixed at 2 mW, and the measurement point was always placed at the center line of the glass capillary tube (GC) or RV. ⁴¹  

Hematocrit variation within physiologic range in the blood samples had little effect on SBR value. When the hematocrit is lower, the effective penetration depth of the laser would probably be deeper, and this would result in little change in the total number of moving blood cells in the sampled volume. The possibility that pathologic hematocrit values affect the SBR, however, cannot be excluded. 

The previous studies indicated that the speckle signals currently quantified as SBR values from a retina vessel were influenced, not only by the blood velocity through it, but also the vessel diameter and speckle signals from the underlying tissues mainly representing choroidal circulation. ^36,37 In the present study, the blood velocity through a major RV in normal, young Japanese was estimated by correcting the raw SBR obtained from the vessel with nomograms that were constructed with a simple model eye and theoretically should be valid only in this simple model eye. Reflectance of human retinal tissue in response to a laser with a 711-nm wavelength was reported to be 7.7%. ³⁹ The absorbance of human retina and choroid in whites and blacks in response to a laser with an 800-nm wavelength is reported to be approximately 5% and 30%, respectively. ³⁸ Based on these reports, the reflectance and absorbance of the film behind a GC as an RV analogue was arbitrarily fixed at 8% and 20% in this model. There was a linear correlation between the SBR obtained from a GC as an RV analogue and the blood velocity through it, its internal diameter, and blood flow rate through posteriorly placed GCs which functioned as a CV analogue. These findings confirmed that the influence of the vessel diameter and choroidal and deep layer retinal circulation should be quantified and corrected to estimate the in vivo blood velocity through a major RV in living eyes with the laser speckle method. 

Because the above relationships among each parameter were linear, nomograms that illustrate the relationships among the actual blood velocity though a surface GC (RV analogue), the SBR values obtained from it, its internal diameter, and the SBR obtained from the posteriorly placed GCs as CV analogue in the model eye could be constructed. 

When the nomograms obtained were directly applied to the results obtained by the laser speckle method in the current normal young Japanese eyes, the estimated blood velocity through a temporal superior retinal artery with a diameter of approximately 115 μm agreed well with that given by bidirectional LDV at the same site of the same artery. The present result was also consistent with the result of a previous study in which LDV was used to measure the blood velocity through human major retinal arteries. ²⁴ The reproducibility of the in vivo blood velocity and flow measurements by the current laser speckle method was in an acceptable range, being slightly better than that for LDV. In addition, the time required for subjects to undergo measurements was shorter for the laser speckle method than for LDV. 

An advantage of the laser speckle method is that it allows blood flow parameters to be measured from both the major retinal artery and accompanying vein (13.7 ± 3.1 mm/s, 11.1 ± 2.2 μL/min, 131 ± 24.1 μm, for blood velocity, flow, and diameter, respectively) at the same time without extending the measurement time by recording a single digital image that includes both vessels. It is, of course, also possible to measure the tissue blood velocity in the optic nerve head or choroid in the same eye. ^29,31,42,43 Analyses of the results were performed on data digitally stored on MO disks by an investigator masked to information about the experiment, which decreased the bias. The present results suggest that the laser speckle method has the potential be used to measure the retinal blood velocity and flow through a major RV under various conditions or pharmacologic influences in human eyes. 

The apparent limitation of the current method is that the in vivo blood velocity must be calculated using nomograms constructed based on the results in a model eye, which is obviously oversimplified. Further, the reflectance and absorbance of the fundus in subject eye must be arbitrarily assumed, although the agreement with the LDV result suggested that the current assumption as the first-order approximation may not be far from the reality as far as normal young Japanese eyes are concerned. If the subject's eyes were old or not normal, such as those with high myopia or retinal disease, or not Japanese, reflectance and absorbance values should be different. Therefore, in clinical studies, it will be necessary to calibrate the results obtained by the laser speckle method once to those by LDV at the same site in the same eye. Once a conversion factor is obtained, however, measurements in the same eye could be more easily performed using the laser speckle method, yielding information about not only blood velocity or flow through major retinal arteries, but also that through accompanying major veins. Peripheral tissue blood velocity in the optic nerve head or choroid can be also obtained using the same instrument. 

Finally, the potential hazards to human eyes of the current measurements must be discussed. The maximum permissible exposure of the retina for viewing a diffuse reflection of a diode laser is 460 mW/cm² for 10 seconds. The maximum exposure of the retina with the present apparatus was approximately 90 mW/cm² for an exposure time of 10 seconds, which is well below the permissible limits. ⁴⁴  

Footnotes

 Disclosure: M. Nagahara, None; Y. Tamaki, None; A. Tomidokoro, None; M. Araie, None