Reliability of different image analysis methods for scanning laser Doppler flowmetry

(1)

Article

Reference

Reliability of different image analysis methods for scanning laser Doppler flowmetry

KREIS, Andréas Josef, et al.

Abstract

To assess reliability and reproducibility of different analysis methods for retinal capillary flow, volume, and velocity from scanning laser Doppler flowmetry (SLDF) topography.

KREIS, Andréas Josef, et al. Reliability of different image analysis methods for scanning laser Doppler flowmetry. Current Eye Research, 2008, vol. 33, no. 5, p. 493-9

DOI : 10.1080/02713680802069149 PMID : 18568887

Available at:

http://archive-ouverte.unige.ch/unige:93229

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=icey20

Download by: [Université de Genève] Date: 06 April 2017, At: 05:26

Current Eye Research

ISSN: 0271-3683 (Print) 1460-2202 (Online) Journal homepage: http://www.tandfonline.com/loi/icey20

Reliability of Different Image Analysis Methods for Scanning Laser Doppler Flowmetry

Andreas J. Kreis, Thanh Nguyen, Sophie Rogers, Jie Jin Wang, Joanna Harazny, Georg Michelson, H. M. Omar Farouque & Tien Y. Wong

To cite this article: Andreas J. Kreis, Thanh Nguyen, Sophie Rogers, Jie Jin Wang, Joanna Harazny, Georg Michelson, H. M. Omar Farouque & Tien Y. Wong (2008) Reliability of Different Image Analysis Methods for Scanning Laser Doppler Flowmetry, Current Eye Research, 33:5-6, 493-499, DOI: 10.1080/02713680802069149

To link to this article: http://dx.doi.org/10.1080/02713680802069149

Published online: 02 Jul 2009.

Submit your article to this journal

Article views: 71

View related articles

(3)

Current Eye Research, 33:493–499, 2008 Copyrightc Informa Healthcare USA, Inc.

ISSN: 0271-3683 print / 1460-2202 online DOI: 10.1080/02713680802069149

Reliability of Different Image Analysis Methods for Scanning Laser Doppler Flowmetry

Andreas J. Kreis, Thanh Nguyen, and Sophie Rogers

Centre for Eye Research Australia, University of Melbourne, Australia

Jie Jin Wang

Centre for Eye Research Australia, University of Melbourne, Australia, and Centre for Vision Research, Westmead Millennium Institute, University of Sydney, Australia

Joanna Harazny and Georg Michelson University Eye Hospital of Erlangen, University of Erlangen, Germany

H. M. Omar Farouque Department of Cardiology, Austin Health Hospital, University of Melbourne, Australia

Tien Y. Wong

Centre for Eye Research Australia, University of Melbourne, Australia, and Singapore Eye Research Institute, Yong Loo Lin School of

Medicine, National University of Singapore, Singapore

ABSTRACT Purpose: To assess reliability and reproducibility of different analy- sis methods for retinal capillary flow, volume, and velocity from scanning laser Doppler flowmetry (SLDF) topography.Material and Methods: SLDF topography analysis using the default retinal flowmeter (HRF) were compared to that using automatic full-field perfusion image analyzer (AFFPIA) and quantified as intraclass correlation coefficients (ICC).Results: The AFFPIA full-field method had the highest reliability, with ICC 0.99 for capillary flow. The reproducibility using the AFFPIA full-field method was high ICC 0.74 for capillary flow.

Conclusions: The AFFPIA full-field method is highly reliable and superior to the default HRF software.

KEYWORDS AFFPIA; flowmetry; image analysis; reliability; reproducibility

INTRODUCTION

The retina can be viewed directly and non-invasively, allowing an assessment of the human microvasculaturein vivo. Recent studies show that structural retinal microvascular changes, such as the presence of retinopathy or variations in retinal vascular caliber, are predictive of a range of systemic cardiovascular and ocular conditions.¹⁻⁹ Development in scanning laser Doppler flowmetry (SLDF) has made quantitative assessment possible of functional changes in retinal blood flow. This technology allows further understanding of physiological and pathological alterations in the retinal microcirculation in patients with cardiovascular disease, diabetic retinopathy, age-related macular degeneration, and glaucoma.¹⁰⁻¹⁴

A commonly used SLDF device is the Heidelberg Retinal Flowmeter (HRF, Heidelberg Engineering, Heidelberg, Germany), which measures retinal capillary flow, volume, and velocity.¹⁵^,¹⁶ While the HRF has been applied to studying different ocular diseases, including glaucoma, diabetic retinopathy, and age- related macular degeneration,¹⁷⁻¹⁹the reliability of measurement captured on the HRF is uncertain.^18,20 The HRF captures a topographic image, but these images require a trained observer to analyze and calculate the summary measurements. There are very little data on the “intra-observer reliability” of such analyses (i.e., the ability of a trained observer or grader to obtain consistent

Received 19 December 2007 Accepted 14 March 2008

Correspondence:Tien Yin Wong, M.D., Ph.D., Centre for Eye Research Australia, University of Melbourne, 32 Gisborne Street, Victoria 3002, Australia. E-mail:

twong@unimelb.edu.au

493

(4)

summary measurements when analyzing the same topographic images of the HRF more than once).²¹Most previous research, with varying terminology, assesses reproducibility (i.e., the intra-subject variation of measurements analyzed from topographs taken over time, where this analysis may be undertaken by one or many trained graders).²²

We aim to assess intra-observer reliability of retinal capillary flow, volume and velocity measurements of the HRF, and intra-subject reproducibility of multiple topographic scans obtained over a short period of time on the same examination day, using four analysis methods: the original default HRF software²³ with cursor options of 10×10 pixels (method 1) or 20×20 pixels (method 2), and the automatic full-field perfusion image analyzer (AFFPIA) with cursor options of 20×20 pixels (method 3) or a full-field option (method 4).¹²

METHODS

We conducted a cross-sectional study of 13 healthy subjects who were volunteers from the administrative and medical staff in the Department of Cardiology at the Austin Health Hospital, University of Melbourne, Victoria, Australia. The Research Ethics Committees of Austin Health, Royal Victorian Eye and Ear Hospital, and the University of Melbourne approved the study protocol. The study was conducted in accordance with the Declaration of Helsinki, and informed written con- sent was obtained from each participant. Eligibility cri- teria were normal visual function and no pathological findings upon slit lamp and fundus examinations. All subjects had not eaten or smoked for at least 3 hr and had rested prior to measurement by a single experienced grader.

Scanning Laser Doppler Flowmetry (SLDF)

The technique of SLDF on the HRF has been de- scribed in detail elsewhere.²⁴^,²⁵ Briefly, the examination was performed in the sitting position after a short rest (about 10–15 min) at room temperature and diffuse natural light, with pupil undilated. Measurements were performed in a chosen region within 1.5 to 2 disc di- ameters temporally from the optic disc margin. An area of 2.56×0.64×0.30 nm was scanned within 2 sec at a resolution of 256 points×64 lines×128 lines with the default 780-nm wavelength laser head installed in the HRF camera.

During the data acquisition, the participant was asked to fixate on a mounted artificial light spot placed approximately 2 m in front of him/her but slightly higher than the level of neutral eye position to prevent blink- ing. A retinal scanning image was taken from the right eye of each person. Repeated images were taken after 5 min and 10 min.

Image Analysis Methods and Modes

We followed the original HRF Operation Manual 1.02 (Heidelberg Engineering).²³ Image analysis was performed using the original default HRF MSDOS- based analysis software (Figure 1 shows 20×20 pixel cursor option) and the AFFPIA (Fig. 2) in accordance with the methods outlined elsewhere.12,15,16,23,25,26

Briefly, AFFPIA calculates the Doppler frequency shift of the hemodynamic flow of each pixel from the en- tire image, while the standard HRF software only an- alyzes within a given small-framed area (i.e., the pixel cursor). For a valid estimation of retinal blood flow, AFFPIA, unlike the HRF software, adjusts brightness to mask under- or over-exposed pixels. AFFPIA also eliminates noise from artificial movement (saccades), avoids measuring extremely wide retinal vessels and the optic nerve head, and accounts for the heart phases (sys- tole and diastole) by averaging the differences between the two phases. Figure 2 shows the steps of imaging processing.¹²

Two different analysis options were utilized from each of two different image analysis software programs, resulting in four methods tested:

1. HRF software, 10 × 10 pixel cursor option (the default mode).

2. HRF software, 20×20 pixel cursor option (Fig. 1).

3. AFFPIA, 20×20 pixel cursor option.

4. AFFPIA, full-field option (Figure 2).

Reliability and Reproducibility

In this study, “intra-observer reliability” refers to comparison of initial and repeated image analysis measurements by one grader (AK), approximately 10 days apart, from the same topographic scanning image. “Intra-subject reproducibility” refers to comparison of measurements from three different scanning topographic images from the same person, taken on the same day 5 min apart (baseline, 5 min later, 10 min later), and analyzed once by the same grader.

A. J. Kreis et al. 494

(5)

FIGURE 1 HRF software, 20×20 pixel cursor option.

Statistical Analysis

All image analysis data were manually exported into a database. Intra-observer reliability and intra-subject reproducibility of multiple topographic images were assessed using intraclass correlation coefficients (ICC) with exact 95% confidence intervals. Reliability utilized per-image data (n =39 images, 3 from each of 13 patients) and data from the first scan per person (n = 13 images), while reproducibility utilized all three images from the same person (n=13 patients). Reliabil-

ity was calculated for all four different analysis modes.

Reproducibility was calculated using only the most reliable image analysis mode. All analyses were undertaken using Intercooled Stata 9.0 for Windows (StataCorp, College Station, TX, USA).

RESULTS

The 13 hospital employee volunteers who partici- pated in this study were predominately female (77%),

FIGURE 2 Automatic full-field perfusion image analyzer (AFFPIA), full-field option: A graphic representation of the image processing procedure of AFFPIA.

495 Reliability of Scanning Laser Doppler Flowmetry

(6)

FIGURE 3 Measurements of capillary flow from the first and repeated image analysis per image using four image analysis methods:

HRF 10×10 cursor (top left), HRF 20×20 cursor (top right), AFFPIA 20×20 cursor (bottom left), and AFFPIA full-field (bottom right).

Abbreviations: AFFPIA, automatic full-field perfusion image analyzer; HRF, standard image analysis software of the Heidelberg retinal flowmeter.

Caucasian (85%), non-smokers (85%), with education qualifications higher than high school (92%). Average age was 39.7 years (range 24–55), and median BMI was 23.9 (70% normal weight, 15% overweight, and 15% obese). One participant (8%) had treated hypertension. No study participant had a history of diabetes, hypercholesterolemia, coronary heart disease, stroke, glaucoma, age-related macular degeneration, diabetic retinopathy, cataract, or retinal arteriovenous occlusion.

Figure 3 shows the retinal capillary flow as analyzed 10 days apart by the four image analysis methods. The AFFPIA full-field method provided near-identical results (bottom right panel).

Table 1 presents intra-observer reliability for capillary flow, volume, and velocity measured using the four different image analysis methods in 39 scans taken from 13 participants (3 scans from each subject). The AFFPIA full-field method gave the greatest reliability for flow (ICC=0.99, 95% CI 0.98, 0.99), volume (ICC=0.99,

TABLE 1 Reliability of the four different image analysis methods in measuring capillary flow, volume, and velocity

Reliability ICC (95% CI)^∗

Flow Volume Velocity

HRF 10×10 cursor 0.43 (0.14, 0.66) 0.63 (0.40, 0.79) 0.41 (0.11, 0.64)

HRF 20×20 cursor 0.62 (0.38, 0.78) 0.78 (0.62, 0.88) 0.62 (0.38, 0.78)

AFFPIA 20×20 cursor 0.77 (0.61, 0.87) 0.86 (0.75, 0.92) 0.76 (0.58, 0.86)

AFFPIA full-field 0.99 (0.98, 0.99) 0.99 (0.98, 0.99) 0.98 (0.96, 0.99)

∗Reliability calculated using intraclass correlation coefficients with exact 95% confidence intervals. Abbreviations: AFFPIA, automatic full-field perfusion image analyzer; HRF, standard image analysis software of the Heidelberg retinal flowmeter.

(7)

TABLE 2 Reproducibility of capillary flow measurements from topographic scans taken at baseline, 5 min, and 10 min later using automatic full-field perfusion image analyzer full-field mode

Reproducibility of flow measurements ICC (95% CI) Over 5 min (baseline vs. 5 min) 0.84 (0.57, 0.95) Over 5 min (5 min vs. 10 min) 0.63 (0.16, 0.87) Over 10 min (baseline vs. 0.74 (0.35, 0.91)

10 min)

Across all three images 0.74 (0.48, 0.90)

95% CI 0.98, 0.99), and velocity (ICC = 0.98, 95%

CI 0.96, 0.99). Image analysis methods using cursor options were less reliable than the AFFPIA full-field method.

Table 2 shows the intra-subject reproducibility of measures (using the AFFPIA full-field method) of multiple images per person, taken 5 min apart (baseline, 5 min, and 10 min later). An overall reproducibility of ICC 0.74 (95% CI 0.48, 0.90) was obtained.

DISCUSSION

The HRF has been used to measure blood flow in different diseases, such as glaucoma, diabetic retinopathy, and age-related macular degeneration among others,¹⁸^,¹⁹as it is well known that disturbances in retinal microvasculature are features of these diseases.⁴^,⁵^,²⁷ Our results show that standard HRF software did not provide highly reliable measures of the SLDF parameters, but newer analysis software utilizing the full field of topographic images improves the reliability substantially. Using the image analysis mode with near- perfect reliability (AFFPIA full-field mode), we found an overall high reproducibility for capillary flow measurements.

Past research into HRF may not be directly comparable to our results due to the use of varied cursor sizes ranging from 2 ×2²⁸ pixels to 50×50 pixels,²⁹ the use of modified HRF devices with different types of laser heads (670 nm,¹⁶default 780 nm,³⁰790 nm¹⁶) and different study designs or statistical methods. Iester and colleagues used AFFPIA software with full-field option, appropriate reliability experimental design, and calculated reliability as a coefficient of variation (CoV) ranging from 3.8–20.7% across different sections of the retina in a sample of healthy and glaucomatous eyes³¹ Using a different statistical approach, our reliability of ICC 0.99 (95% CI 0.98–0.99) for capillary flow mea-

sured using AFFPIA full-field in healthy eyes is suffi- ciently high for repeated measures by the same observer, although this estimate is not directly comparable to the CoV values reported by other studies.

Many researchers have assessed reproducibility of results across subjects over time.¹²^,²⁶^,³²⁻³⁵In human subjects using HRF software, capillary flow reproducibility estimates have been reported (for the 10 × 10 pixel cursor) in the range of 0.82 (CoV)¹⁶ to 0.64 (ICC),³⁶ where images were taken one day apart, and 91.4%

(CoV)³⁷ for images taken within 10 min. Using AFF- PIA software with the full-field option, reproducibility estimates for capillary flow have been reported in the range of 0.4–1.9% (CoV)³⁸for study designs where images were taken days apart, and 0.8–0.95 (ICC)³⁵ for images taken within 30 min.³⁵^,³⁸Our results for reproducibility of ICC 0.74 over 5 and 10 min using AFFPIA full-field were close to those previous reports of ICC 0.8–0.95.

Reliability is affected by different image analysis factors, such as the choice of and size of measurement area and quality and clarity of the topographic image.²⁹ Since the HRF takes 2 sec to capture a topographic image, working from the top left to the bottom right,²³^,³⁹ the resultant image is subject to spatial heterogeneity in circulatory parameters due to heart beat-associated pulsations.¹⁶ As such, the placement of the cursor, or the choice and size of image area to be included in analysis, becomes important.²⁹Griesser and colleagues showed that the variation in HRF parameters decreased as the cursor size increased.³⁷The AFFPIA full-field option provides the highest reliability as it includes the largest possible image area for analysis, and automati- cally adjusts for the quality of the topographic image,

“auto-cleaning” and masking any extra-wide vessels, optic nerve head, and over- or under-exposed pixels.¹²So, for a given cursor size, we would expect the AFFPIA reliability to be better than for HRF software. All image analysis issues that affect reliability will affect reproducibility as well as imaging-related issues, such as choice of location within the eye itself, distance between camera and eye, room lighting, sharpness of topographic image, and changes within the subject, such as level of relaxation¹⁴ and ability to concentrate and fixate.¹⁴^,²⁸^,³³^,⁴⁰^,⁴¹ The effect these imaging-related factors have on reproducibility will likely increase as the time between images increases.

The strengths of this study include the use of healthy subjects, with no eye pathology, who had not eaten

(8)

or smoked for at least 3 hr and who rested prior to measurement by a single experienced grader. Possible limitations of the study include the small sample size (n =13), the measurements not synchronized on the cardiac cycle, and the use of diffuse natural lighting rather than constant artificial lighting.¹⁴

In conclusion, we found that default HRF software, using a cursor-based method, does not provide reliable measures of capillary circulation parameters from HRF topographs. Newer image analysis software (AFFPIA) utilizing the full field of topographic images improves the reliability substantially and provides high intra- observer reliability for repeated measures on the same images, and high intra-subject reproducibility for measures on different images taken from the same subject over a short period of time. Therefore, we recommend using the AFFPIA full-field mode of image analysis in conjunction with the HRF camera for future research.

Since SLDF is a relatively new technique, correlations between SLDF-measured capillary circulatory parameters and biologically plausible factors need to be assessed.

REFERENCES

[1] Wang JJ, Liew G, Klein R, et al. Retinal vessel diameter and cardiovascular mortality: Pooled data analysis from two older populations.

Eur Heart J.2007;28:1984–1992.

[2] Liew G, Kaushik S, Rochtchina E, Tan AG, Mitchell P, Wang JJ. Retinal vessel signs and 10-year incident age-related maculopathy: The Blue Mountains Eye Study.Ophthalmology.2006;113:1481–1487.

[3] Mitchell P, Leung H, Wang JJ, et al. Retinal vessel diameter and open- angle glaucoma: The Blue Mountains Eye Study.Ophthalmology.

2005;112:245–250.

[4] Wong TY, Wong T, Mitchell P. The eye in hypertension.Lancet.

2007;369:425–435.

[5] Nguyen TT, Wang JJ, Wong TY. Retinal vascular changes in pre- diabetes and prehypertension: New findings and their research and clinical implications.Diabetes Care.(Epub Jun 2007;30:2708–2715).

[6] Wong TY, Klein R, Sharrett AR, et al. Retinal arteriolar narrowing and risk of coronary heart disease in men and women. The Atheroscle- rosis Risk in Communities Study.JAMA.2002;287:1153–1159.

[7] Wong TY, Klein R, Sharrett AR, et al. Retinal arteriolar narrowing and risk of diabetes mellitus in middle-aged persons.JAMA.

2002;287:2528–2533.

[8] Wong TY, Rosamond W, Chang PP, et al. Retinopathy and risk of congestive heart failure.JAMA.2005;293:63–69.

[9] Wong TY, Klein R, Couper DJ, et al. Retinal microvascular abnormal- ities and incident stroke: The Atherosclerosis Risk in Communities Study.Lancet.2001;358:1134–1140.

[10] Michelson G, Warntges S, Baleanu D, et al. Morphometric age- related evaluation of small retinal vessels by scanning laser Doppler flowmetry: Determination of a vessel wall index. Retina.

2007;27:490–498.

[11] Werner D, Michalk F, Harazny J, Hugo C, Daniel WG, Michelson G.

Accelerated reperfusion of poorly perfused retinal areas in central retinal artery occlusion and branch retinal artery occlusion after a short treatment with enhanced external counterpulsation.Retina.

2004;24:541–547.

[12] Michelson G, Welzenbach J, Pal I, Harazny J. Automatic full-field analysis of perfusion images gained by scanning laser Doppler flowmetry.Br J Ophthalmol.1998;82:1294–1300.

[13] Michelson G, Harazny J. Increased vascular resistance for ve- nous outflow in central retinal vein occlusion. Ophthalmology.

1997;104:659–663.

[14] Harazny JM, Ritt M, Baleanu D, et al. Increased wall:lumen ratio of retinal arterioles in male patients with a history of a cerebrovascular event.Hypertension.2007;50:623–629.

[15] Zinser G.Scanning laser Doppler flowmetry. Principle and technique.

The Hague: Kugler Publications, 1999:197–204.

[16] Michelson G, Schmauss B, Langhans MJ, Harazny J, Groh MJ. Prin- ciple, validity, and reliability of scanning laser Doppler flowmetry.J Glaucoma.1996;5:99–105.

[17] Nicolela MT, Hnik P, Drance SM. Scanning laser Doppler flowmeter study of retinal and optic disk blood flow in glaucomatous patients.

Am J Ophthalmol.1996;122:775–783.

[18] Michelson G, Langhans MJ, Groh MJ. Clinical investigation of the combination of a scanning laser ophthalmoscope and laser Doppler flowmeter.Ger J Ophthalmol.1995;4:342–349.

[19] Nowak I, Pilas-Pomykalska M, Czajkowski J, Omulecki W. [Retinal capillary blood flow in exudative age-related macular degeneration].

Klin Oczna.2007;109:135–137.

[20] Mavroudis L, Harris A, Topouzis F, et al. Reproducibility of pixel-by- pixel analysis of Heidelberg retinal flowmetry images: The Thessa- loniki Eye Study.Acta Ophthalmol Scand.Sept. 10, 2007.

[21] Institute of Electrical and Electronics Engineers (IEEE).IEEE Standard Computer Dictionary: A Compilation of IEEE Standard Computer Glossaries. New York: IEEE; 1990.

[22] IUPAC. Definition of Reproducibilty.IUPAC Compendium of Chemi- cal Terminology.North Carolina: IUPAC. 1994;66:598.

[23] Heidelberg Engineering. Heidelberg Retina Flowmeter, Operation Software Release 1.02.1.02.

[24] Michelson G, Groh M, Langhans M, Schmauss B. [2-Dimensional mapping and retinal and papillary microcirculation using scanning laser Doppler flowmetry]. Klin Monatsbl Augenheilkd.

1995;207:180–190.

[25] Bonner RF, Nossal R.Principles of Laser Doppler Flowmetry. Boston:

Kluwer; 1990.

[26] Michelson G, Welzenbach J, Pal I, Harazny J. Functional imaging of the retinal microvasculature by scanning laser Doppler flowmetry.

Int Ophthalmol.2001;23:327–335.

[27] Wong TY, Kamineni A, Klein R, et al. Quantitative retinal venu- lar caliber and risk of cardiovascular disease in older persons:

The cardiovascular health study.Arch Intern Med.2006;166:2388–

2394.

[28] Chauhan BC, Yu PK, Cringle SJ, Yu DY. Confocal scanning laser Doppler flowmetry in the rat retina: Origin of flow signals and dependence on scan depth. Arch Ophthalmol. 2006;124:397–

402.

[29] Bohdanecka Z, Orgul S, Prunte C, Flammer J. Influence of acquisition parameters on hemodynamic measurements with the Heidel- berg Retina Flowmeter at the optic disc.J Glaucoma.1998;7:151–

157.

[30] Yu DY, Townsend R, Cringle SJ, Chauhan BC, Morgan WH. Improved interpretation of flow maps obtained by scanning laser Doppler flowmetry using a rat model of retinal artery occlusion.Invest Oph- thalmol Vis Sci.2005;46:166–174.

[31] Iester M, Altieri M, Michelson G, Vittone P, Calabria G, Traverso CE. Intra-observer reproducibility of a two-dimensional mapping of the optic nerve head perfusion. J Glaucoma. 2002;11:488–

492.

[32] Jonescu-Cuypers CP, Harris A, Wilson R, et al. Reproducibility of the Heidelberg retinal flowmeter in determining low perfusion areas in peripapillary retina.Br J Ophthalmol.2004;88:1266–1269.

[33] Kagemann L, Harris A, Chung HS, Evans D, Buck S, Martin B. Heidel- berg retinal flowmetry: Factors affecting blood flow measurement.

Br J Ophthalmol.1998;82:131–136.

(9)

[34] Nicolela MT, Hnik P, Schulzer M, Drance SM. Reproducibility of retinal and optic nerve head blood flow measurements with scanning laser Doppler flowmetry.J Glaucoma.1997;6:157–164.

[35] Hafez AS, Bizzarro RL, Rivard M, et al. Reproducibility of retinal and optic nerve head perfusion measurements using scanning laser Doppler flowmetry.Ophthalmic Surg Laser Imag.2003;34:422–432.

[36] Jonescu-Cuypers CP, Harris A, Bartz-Schmidt KU, et al. Repro- ducibility of circadian retinal and optic nerve head blood flow measurements by Heidelberg retina flowmetry.Br J Ophthalmol.

2004;88:348–353.

[37] Griesser SM, Lietz A, Orgul S, et al. Heidelberg retina flowmeter parameters at the papilla in healthy subjects.Eur J Ophthalmol.

1999;9:32–36.

[38] Iester M, Ciancaglini M, Rolle T, Vattovani O. Observer interpretation variability of peripapillary flow using the Heidelberg Retina Flowme- ter.Eye.2006;20:1246–1253.

[39] Michelson G, Schmauss B. Two-dimensional mapping of the perfusion of the retina and optic nerve head. Br J Ophthalmol.

1995;79:1126–1132.

[40] Tsang AC, Harris A, Kagemann L, Chung HS, Snook BM, Garzozi HJ.

Brightness alters Heidelberg retinal flowmeter measurements in an in vitro model.Invest Ophthalmol Vis Sci.1999;40:795–799.

[41] Hosking SL, Embleton S, Kagemann L, Chabra A, Jonescu-Cuypers C, Harris A. Detector sensitivity influences blood flow sampling in scanning laser Doppler flowmetry.Graefes Arch Clin Exp Ophthal- mol.2001;239:407–410.