Upload
aljoscha-s-neubauer
View
218
Download
2
Embed Size (px)
Citation preview
Introduction
Diabetic retinopathy (DR) is a leadingcause of visual impairment and blind-ness in developed countries. The inci-dence and prevalence are high andincreasing (Stefansson et al. 2000): itis estimated that 10.2 million USadults aged ‡ 40 years currently havediabetes mellitus (DM). The crudeprevalence rates for retinopathy andvision-threatening retinopathy are40.3% and 8.2% (Kempen et al.2004). Macular oedema is a majorcause of vision loss in diabetespatients and proliferative retinopathyis another (Klein et al. 1984). Anannual clinical examination withdilated pupils by an ophthalmologistis the current gold standard of carefor reducing diabetes-related ocularcomplications in patients with no ormild DR (Fong et al. 2004). In theEarly Treatment Diabetic RetinopathyStudy (ETDRS 1985) this, combinedwith timely treatment, led to a 50%reduction in moderate visual loss.
Good documentation of DR – oftenreferred to as a reference standard – isprovided by 7-field stereoscopic 30-degree fundus photography as definedby the ETDRS (1991). However, thisdocumentation procedure usuallyrequires expensive equipment and askilled photographer and therefore isusually applied mostly in clinical
Digital fundus imagegrading with the non-mydriaticVisucam
PRO NMversus the
FF450plus camera in diabeticretinopathy
Aljoscha S. Neubauer,1 Antje Rothschuh,2 Michael W. Ulbig1 andMarcus Blum2
1Department of Ophthalmology, Ludwig-Maximilians-University, Munich, Germany2Department of Ophthalmology, Helios Klinikum Erfurt GmbH, Erfurt, Germany
ABSTRACT.
Purpose: Grading diabetic retinopathy in clinical trials is frequently based on
7-field stereo photography of the fundus in diagnostic mydriasis. In terms of
image quality, the FF450plus camera (Carl Zeiss Meditec AG, Jena, Germany)
defines a high-quality reference. The aim of the study was to investigate if the
fully digital fundus camera VisucamPRO NM could serve as an alternative in
clinical trials requiring 7-field stereo photography.
Methods: A total of 128 eyes of diabetes patients were enrolled in the random-
ized, controlled, prospective trial. Seven-field stereo photography was per-
formed with the VisucamPRO NM and the FF450plus camera, in random order,
both in diagnostic mydriasis. The resulting 256 image sets from the two cam-
era systems were graded for retinopathy levels and image quality (on a scale
of 1–5); both were anonymized and blinded to the image source.
Results: On FF450plus stereoscopic imaging, 20% of the patients had no or
mild diabetic retinopathy (ETDRS level £ 20) and 29% had no macular
oedema. No patient had to be excluded as a result of image quality. Retinopa-
thy level did not influence the quality of grading or of images. Excellent over-
all correspondence was obtained between the two fundus cameras regarding
retinopathy levels (j 0.87) and macular oedema (j 0.80). In diagnostic mydria-
sis the image quality of the Visucam was graded slightly as better than that of
the FF450plus (2.20 versus 2.41; p < 0.001), especially for pupils < 7 mm in
mydriasis.
Conclusions: The non-mydriatic VisucamPRO NM offers good image quality
and is suitable as a more cost-efficient and easy-to-operate camera for applica-
tions and clinical trials requiring 7-field stereo photography.
Key words: diabetes – diabetic retinopathy – fundus photography – imaging – retinal screening
Acta Ophthalmol. 2008: 86: 177–182Copyright ª Acta Ophthalmol Scand 2007.
doi: 10.1111/j.1600-0420.2007.01029.x
Acta Ophthalmologica 2008
177
trials. Recently, digital photographyinstead of 35-mm slide film was ableto facilitate photographic documenta-tion without loss in sensitivity andspecificity (George et al. 1997, 1998;Henricsson et al. 2000; Bursell et al.2001; Sharp et al. 2003). However,acquiring 7-field stereoscopic fundusphotographs remains a difficult andexpensive procedure despite the capac-ity for digital image storage as itrequires high-quality fundus camerasand a trained photographer. Othertechnologies or non-mydriatic cameraswith good diagnostic properties canbe used for screening purposes (Neu-bauer et al. 2003). In Europe, theEURODIAB standard is often appliedfor screening. This is based on two45-degree photographs, one centredon the macula and one centred nasallyto the optic disc (Aldington et al.1995; Gibbins et al. 1998). For clinicaltrials such as those involving pharma-ceutical products or trials to docu-ment DR, 7-field stereo photographyis still considered to be a well estab-lished reference standard and there-fore was used in our study.
Recently, a novel non-mydriaticfundus camera, the fully digital Visu-camPRO NM (Carl Zeiss Meditec AG,Jena, Germany), has become avail-able. It is much less costly and easierto operate than a current industrystandard for high-quality mydriaticfundus imaging, the Zeiss FF450plus
camera (Carl Zeiss Meditec AG). Inan effort to facilitate 7-field ETDRSphotography and make it more cost-efficient (Martin & Yidegiligne 1998)and available as a reference, especiallyfor clinical trials, we investigated theVisucamPRO NM for mydriatic 7-fieldstereo photography in DR. Gradingresults and image quality were com-pared with the latest mydriatic onlyFF450plus camera, which served as areference standard.
Methods
Patients
Consecutive patients were recruited atthe outpatient clinic of the Depart-ment of Ophthalmology, Helios Klini-kum Erfurt GmbH. Patients aged‡ 18 years with either type 1 or type 2diabetes (based on World HealthOrganization criteria, WHO 2001)were included after they had given
informed consent. A total of 128 eyesfrom 64 patients were included. Thestudy received institutional boardapproval and conformed to the princi-ples expressed in the Declaration ofHelsinki. Exclusion criteria were preg-nancy or lactation and a history ofconditions in either eye that mightpreclude pupil dilatation or use ofeyedrops. Furthermore, eyes with eyediseases other than DR or significantmedia opacities preventing adequatefunduscopy and photography wereexcluded. Eyes were also excluded ifthe pupil diameter in diagnosticmydriasis (1% tropicamide and 5%neosynephrine) was < 6 mm. Pupildiameter after pupil dilation was mea-sured on the calibrated millimetrescale of a slit-lamp. No eyes had to beexcluded for any of the exclusion cri-teria, including imaging quality. Botheyes of each eligible patient wereincluded. The mean patient age was60 ± 12 years. A total of 62% weremale, with an overall mean body massindex of 29.8 ± 5 kg ⁄m2. Blood pres-sure (BP) was controlled: mean sys-tolic BP was 136.4 ± 17.6 mmHgand mean diastolic BP was80.3 ± 8.9 mmHg. A total of 86% ofpatients had type 2 diabetes; overallHbA1c was 7.5 ± 1.4%.
Seven-field stereo photography
Standardized 7-field stereo photogra-phy was performed using the non-mydriatic VisucamPRO NM and theFF450plus camera in random order,both in diagnostic mydriasis. Theseven 30-degree stereo fields usedcomplied with the Early Treatment ofDiabetic Retinopathy Protocol (1991)and its procedures as defined by theWisconsin Reading Center (see http://eyephoto.ophth.wisc.edu).
The FF450plus camera allows 20-,30- and 50-degree fundus imaging tovarious photographic devices. Becauseof its optical design, a pupil diameter‡ 5.5 mm is recommended by themanufacturer. A 5.0-megapixel chargecoupled device (CCD) camera (Sony3CCD; Sony, Tokyo, Japan) was uti-lized in this study. Focusing andalignment of the image were per-formed using the ocular tube of thecamera in addition to a previewingcamera. Figure 1 shows the set-up inuse. The non-mydriatic VisucamPRO NM
allows 30- and 45-degree fundus
imaging. The VisucamPRO NM istargeted at less experienced operatorsand, unlike the FF450plus, its focusingand positioning are guided electroni-cally. Infrared illumination helps toadjust the settings before the digitalimage is acquired (Fig. 1). A 5.0-megapixel CCD camera is integratedand was used in the study. With bothcamera systems, the seven 30-degree�stereo images were taken and saveduncompressed in Visupac 4.2 software(Carl Zeiss Meditec AG). The result-ing 14 images per camera systemand patient eye were anonymized andexported in the digital imaging andcommunication in medicine (DICOM)format and transferred on DVD-Rmedia to the reading centre.
Grading
The anonymized datasets from differ-ent patients and the two cameras wereDICOM imported on the Visupacworkstation at the reading centre [Lud-wig-Maximilians-Universitat (LMU),Munich, Germany]. Viewing was per-formed on a colour-calibrated 21-inchcathode ray tube (CRT) monitor(model 21P4; Fujitsu-Siemens,Munich, Germany). The Visupac 4.2stereo display mode was used to deter-mine macular oedema using a simplestereo viewer (Carl Zeiss MeditecAG), so that the two eyes were opti-cally separated when the two imageswere displayed side-by-side on themonitor. The grading of DR adheredto the ETDRS (1991) levels andreferred to standard photographs anddefinitions, where levels 10, 14, 15 and20 correspond to no and mild non-proliferative retinopathy (NPDR), lev-els 35–53 indicate moderate to severeNPRD, and levels 61–85 indicate pro-liferative diabetic retinopathy (PDR).Additionally, macular oedema wasgraded on two grading scales. The firstof these noted the presence or not ofclinically significant macular oedema(CSME) according to by the ETDRS(1985). The second used the 4-level,international clinical diabetic macularoedema disease severity scale (Wilkin-son et al. 2003), with definitions of nomacular oedema, and mild, moderateand severe macular oedema.
Image quality of all images wasgraded on a scale of 1)5 as proposedand validated by Hansen et al. (2004),where: 1 = ‘excellent’ and shows clear
Acta Ophthalmologica 2008
178
details; 2 = ‘good’ and shows cleardetails in central images and < 1 ⁄ 8of peripheral images show no details;3 = ‘acceptable’, with 1 ⁄ 6 of centraland 1 ⁄8 to 1 ⁄ 4 of peripheral imagesshowing no details; 4 = ‘weak’, wherethe central 1 ⁄ 6 to 2 ⁄ 3 and peripheral1 ⁄4 to 2 ⁄ 3 of the image show nodetails; 5 = ‘not gradable’, where> 2 ⁄ 3 of an image shows no details.The macula and optic disc centredimages were considered as ‘central’,whereas the other fields were consid-ered as ‘peripheral’. All grading andimage quality assessment of the ran-domized image sets were performedby one experienced retina specialist(ASN). To finally determine intra-grader agreement of DR grading (i.e.reproducibility of gradings), a subsetof 40 eyes was graded twice with aninterval of 2 months between the gra-dings. In total, 1792 stereo images(128 eyes · 2 cameras · 7 fields) weregraded.
Statistics
All data were collected in a MS-Excel2000 spreadsheet (Microsoft Corpora-tion, Redmond, WA, USA) and anal-ysed using SPSS 13.0 for Windows(SPSS Inc., Chicago, IL, USA).Before the study, power analysis wasperformed to show the equivalence ofboth camera systems for grading pur-poses. It yielded (with a = 5%) a nec-essary n = 39 patients for a statisticalpower of 80% (and n = 55 for 90%power) for a chosen d of 0.20. Besidessensitivity and specificity (including
their 95% confidence intervals, CI),the analysis focused on weighted jstatistics, which were calculated andassessed as proposed in (Altman1991):
(1) poor agreement: j < 0.20,(2) fair agreement: j = 0.20)0.40,(3) moderate agreement: j = 0.41)0.60,(4) good agreement: j = 0.61)0.80,(5) very good agreement: j = 0.81)1.00.
Weighting of jwas performed analo-gous to the Diabetes Control andComplications Trial (1995), so that:0 = differences of > 2 levels; 0.5 =differences of £ 2 levels; 0.75 = dif-ferences of £ 1 level, and 1 = perfectagreement. On all tests p < 0.05 wasconsidered significant. For all compar-isons of quality, non-parametric test-ing (Wilcoxon signed rank test andWilcoxon rank sum test) was applied.
Results
Grading of diabetic retinopathy
Of the 128 eyes, 20% had no or onlymild DR (ETDRS level £ 20) asgraded on the FF450plus camera sys-tem. Grading the image sets obtainedfrom the VisucamPRO NM camera sys-tem resulted in very good correspon-dence with the FF450plus, yielding a jof 0.87 (confidence interval [CI]0.81–0.92). The detailed gradingresults [categorized into internationalclinical diabetic disease severity cate-gories (Wilkinson et al. 2003)] are
summarized in Table 1. The sensitivityof the VisucamPRO NM to correctlyidentify ETDRS levels £ 35 or > 35for significant DR was 99% (CI 94–100%) with a specificity of 92% (CI73–99%). Table 1 shows that differen-tiating no (see first column) versussome fundus changes by DR was alsoperformed well by grading of Visu-camPRO NM images: for ETDRS levelsof < 20 a sensitivity of 99% (CI 97–100%) at a specificity of 86% (CI 74–90%) was obtained. Overall intragrad-er agreement indicating reproducibilityof grading ETDRS levels was good,with j = 0.77 (CI 0.64–0.89).
Grading of macular oedema
Of the 128 eyes, 71% had some degreeof macular oedema as graded on theFF450plus camera system. Grading ofthe image sets from the VisucamPRO NM
camera system resulted in goodcorrespondence, with j = 0.80 (CI0.73–0.87). The detailed grading resultsare summarized in Table 2. Forinstance, the table shows that differen-tiating no (see first column) versussome macular oedema (other columns)could be performed well (sensitivity95% [CI 90–97%], specificity 84% [CI74–90%]). Reproducibility was good,with j = 0.60 (CI 0.42–0.79) for intra-grader agreement. Regarding CSME,41% of all eyes were found to haveCSME on FF450plus camera imaging.Compared with this, the VisucamPRONM
yielded an excellent sensitivity of 91%(CI 79–97%) and specificity of 0.80(CI 0.69–0.88).
Operating inputsDisplay
VISUCAMPRO NM Camera
Digital camera sensor
FF450PLUS camera
Camera unit
Viewing and focusing ocular
Fig. 1. The mydriatic only FF450plus fundus camera and the non-mydriatic VisucamPRO NM camera as used in the study. A 5-megapixel charge
coupled device sensor was used on both cameras.
Acta Ophthalmologica 2008
179
Image quality
Mean image quality on the 5-levelscale was 2.20 for the VisucamPRO NM
and 2.41 for the FF450plus, yieldingsome advantage for the VisucamPRO NM
(p < 0.001). The detailed imagequality distribution is given in Fig. 2.
The ETDRS level of the eye did notinfluence image quality, whereashigher patient age showed some non-significant decrease on image quality(FF450plus: correlation coefficientr = 1.0, p = 0.08; VisucamPRO NM:correlation coefficient r = 1.0, p =
0.33), despite imaging in mydriasis.Pupil diameters in mydriasis andage were non-significantly correlated(r = ) 0.85, p = 0.33). A factorinfluencing image quality was thepresence of cataracts for the FF450plus
(p = 0.048), but the VisucamPRO NM
was less affected (p = 0.07). A smallpupil diameter (6.0–7.0 mm in mydria-sis) degraded the image quality of theFF450plus (p = 0.003) compared withthe VisucamPRO NM, whereas for alllarger pupil diameters > 7.0 mm nosignificant difference in image qualityexisted between the two camera sys-tems. Table 3 shows the dependenceon pupil diameter in detail.
Discussion
In this study, we were able to showthat for grading DR using 7-field ste-reo photography, the fully digital fun-dus camera VisucamPRO NM is at leastcomparable with the Zeiss FF450plus,a current industry standard forhigh-quality, mydriatic fundus pho-tography. The image quality obtainedwith the VisucamPRO NM was in factsuperior for smaller pupil sizes as theVisucamPRO NM requires less pupildiameter because it is designed fornon-mydriatic photography.
The seven stereo photographs asdefined by the ETDRS are known tooffer excellent test characteristics forboth DR and macular oedema (Kin-youn et al. 1989). However, such pho-tography requires a highly skilledphotographer, and is costly and time-consuming, and is therefore not ideal
Table 1. Early Treatment Diabetic Retinopathy Study and International Classification Level Grading of the two fundus cameras.
Number of eyes
FF450Plus
No DR
NPDR
PDR
No
classificationQuestionable Mild Moderate Severe
VisucamPRO NM ETDRS level 10 14 ⁄ 15 20 35
43 ⁄ 4753 61 ⁄ 65
71 ⁄ 7581 ⁄ 85
90
No DR 10 7 0 0 0 0 0 0
NPDR
Questionable 14 ⁄ 15 3 9 1 0 0 0 0
Mild 20 0 2 2 1 0 0 0
Moderate 35 ⁄ 43 ⁄ 47 0 1 1 45 2 0 0
Severe 53 0 0 0 6 22 2 0
PDR 61 ⁄ 65 ⁄ 71 ⁄ 75 ⁄ 81 ⁄ 85 0 0 0 2 1 21 0
No classification 90 0 0 0 0 0 0 0
ETDRS = Early Treatment Diabetic Retinopathy Study; DR = diabetic retinopathy; PDR = proliferative diabetic retinopathy; NPDR = non-
proliferative retinopathy.
Table 2. Macular oedema grading of the two fundus cameras.
Macular oedema
FF450plus
Macular oedemaNo
classificationNone Mild Moderate Severe
VisucamPRO NM Macular oedema
None 31 4 1 0 0
Mild 6 33 6 0 0
Moderate 0 5 18 3 0
Severe 0 1 2 18 0
No classification 0 0 0 0 0
0
10
20
30
40
50
60
70
80
Excellen
t
Go
od
Ac
ce
pta
ble
Po
or
Un
gra
deab
le
Grading
Nu
mb
er
of
eyes
VISUCAM
FF450
Fig. 2. Comparison of image quality of the two camera systems. Overall, a small advantage was
found for the VisucamPRO NM (p < 0.001).
Acta Ophthalmologica 2008
180
for primary screening and is mostlyapplied in clinical trials. It currentlydefines a reference standard for gooddocumentation of DR when per-formed in mydriasis with a high- qual-ity camera such as the FF450plus.Digital photography rather than slidesmight facilitate the procedure to somedegree (George et al. 1997, 1998; Hen-ricsson et al. 2000; Bursell et al. 2001;Sharp et al. 2003), but the equipmentused is still costly and difficult tooperate. Digital guidance of the wholeimaging procedure with focusing andillumination – as provided with theVisucamPRO NM – may help to furtherfacilitate the process. We were able toshow that grading results obtainedwith the Visucam for both DR level(j = 0.87) and macular oedema(j = 0.80) are valid compared withthe FF450plus, which served as a refer-ence standard. The correlationbetween the two systems as well asintragrader reproducibility (j = 0.77and 0.60) was similar to the j valuesobtained in the ETDRS (1991) study.It should be also remembered that thepatient group imaged in the currentstudy included a relatively high per-centage of subjects with advancedstages of DR (Table 1), which makesexact grading more difficult becauseof the many disease-related changes.This is because we selected ourpatients from a diabetes eye clinic,where more advanced retinopathystages are treated. For many otherpurposes, such as screening and fol-lowing lower levels of DR where lowlevels of disease-related changes arepresent, even better characteristics canbe expected. This is supported by theexcellent ability the VisucamPRO NM
to detect retinopathy levels, with asensitivity of 99% and specificity of92% for levels with a cut-off ofETDRS level 35 and similar results fora cut-off of level 20. These diagnosticproperties also fulfil the definitions of,
for instance, the UK’s NationalScreening Committee (http://www.nscretinopathy.org.uk).
The seven EDTRS photographsapplied in our study cover 75–65degrees of the central retina and arecurrently mostly used for clinicaltrials. There has been much investiga-tion to determine whether simplerprotocols are sufficient for screeningpurposes. Although one 60-degreenon-stereoscopic fundus image wasfound insufficiently sensitive (Molleret al. 2001), two 60-degree fundusphotographs (one centred on the mac-ula and one on the optic disc) offergood test characteristics (von Wendtet al. 2000). Two 45-degree� fundusphotographs (one macula- and onedisc-centred) also provide goodscreening characteristics (Stellingwerfet al. 2001; von Wendt et al. 2002).The frequently applied EURODIABstandard is also based on two45-degree photographs, one centredon the macula and the other centrednasal to the optic disc (Aldingtonet al. 1995; Gibbins et al. 1998), andthe US Joslin Vision Network usesthree 45-degree non-mydriatic images(Cavallerano et al. 2003; Perrieret al. 2003). Any of these set-ups aretheoretically possible with the non-mydriatic VisucamPRO NM, which inthis study was shown to offer excellentcharacteristics for mydriatic 7-fieldstereo photography, and which mayserve as a reference standard for tech-nology assessment and clinical trials.However, it should be kept in mindthat stereoscopic photography islimited in terms of its ability to detectmacular thickening (Knudsen &Skriver 2006), and other methods,especially optical coherence tomogra-phy (Neubauer et al. 2003), are muchmore sensitive (Massin et al. 2006).
In summary, the digital non-mydri-atic digital VisucamPRO NM offersimaging and grading characteristics
similar to those of the mydriatic ZeissFF450plus. It is therefore very suitableas a more cost-efficient and easy-to-operate camera for applications andespecially for clinical trials thatrequire 7-field stereo photographs asvalidated here for diabetic retino-pathy.
Acknowledgments
The authors thank Dr ChristophRussmann, Carl Zeiss Meditec AGfor providing valuable technical infor-mation and help with statistics, andAnno Hermanowski, Carl Zeiss Medi-tec AG for technical support. Thestudy was conducted with materialand technical support from Carl ZeissMeditec AG, Jena, Germany. ASNand AR are joint first authors.
ReferencesAldington SJ, Kohner EM, Meuer S, Klein
R & Sjolie AK (1995): Methodology for
retinal photography and assessment of
diabetic retinopathy: the EURODIAB
IDDM complications study. Diabetologia
38: 437–444.
Altman DG (1991): Practical Statistics for
Medical Research. London, UK: Chapman
and Hall, 404.
Bursell SE, Cavallerano JD, Cavallerano AA,
Clermont AC, Birkmire-Peters D, Aiello
LP & Aiello LM (2001): Stereo non-mydri-
atic digital-video colour retinal imaging
compared with Early Treatment Diabetic
Retinopathy Study seven standard field 35-
mm stereo colour photos for determining
level of diabetic retinopathy. Ophthalmol-
ogy 108: 572–585.
Cavallerano AA, Cavallerano JD, Katalinic
P, Tolson AM, Aiello LP & Aiello LM
(2003): Use of Joslin Vision Network digi-
tal-video non-mydriatic retinal imaging to
assess diabetic retinopathy in a clinical pro-
gramme. Retina 23: 215–223.
Diabetes Control and Complications Trial
Research Group (1995): Progression of ret-
inopathy with intensive versus conventional
treatment in the Diabetes Control and
Complications Trial. Ophthalmology 102:
647–661.
Early Treatment Diabetic Retinopathy Study
Research Group (1985): Photocoagulation
for diabetic macular oedema. ETDRS
Report No. 1. Arch Ophthalmol 103:
1796–1806.
Early Treatment Diabetic Retinopathy Study
Research Group (1991): Grading diabetic
retinopathy from stereoscopic colour
fundus photographs – an extension of
the modified Airlie House classification.
Table 3. Image quality grading dependent on pupil diameter in diagnostic mydriasis.
Pupil diameter
(mm)
Number
of eyes
Quality grading (mean ± SD)Statistical significance
for differenceFF450Plus VisucamPRO NM
< 7.0 37 2.6 ± 0.8 2.2 ± 0.6 p = 0.003
7.0–7.9 58 2.3 ± 0.7 2.2 ± 0.7 NS
8.0–8.9 29 2.3 ± 0.6 2.3 ± 0.6 NS
‡ 9.0 4 2.3 ± 1.0 1.5 ± 1.0 NS
SD = standard deviation; NS = not significant.
Acta Ophthalmologica 2008
181
ETDRS Report No. 10. Ophthalmology
98: 786–806.
Fong DS, Aiello L, Gardner TW, King GL,
Blankenship G, Cavallerano JD, Ferris FL
III & Klein R (2004): Retinopathy in dia-
betes. Diabetes Care 27(Suppl. 1): 84–87.
George LD, Leverton C, Young S, Lusty J,
Dunstan FD & Owens DR (1997): Can
digitised colour 35-mm transparencies be
used to diagnose diabetic retinopathy? Dia-
bet Med 14: 970–973.
George LD, Halliwell M, Hill R, Aldington
SJ, Lusty J, Dunstan F & Owens DR
(1998): A comparison of digital retinal
images and 35-mm colour transparencies in
detecting and grading diabetic retinopathy.
Diabet Med 15: 250–253.
Gibbins RL, Owens DR, Allen JC & East-
man L (1998): Practical application of the
European Field Guide in screening for dia-
betic retinopathy by using ophthalmoscopy
and 35-mm retinal slides. Diabetologia 41:
59–64.
Hansen AB, Sander B, Larsen M, Kleener J,
Borch-Johnsen K, Klein R & Lund-Ander-
sen H (2004): Screening for diabetic reti-
nopathy using a digital non-mydriatic
camera compared with standard 35-mm
stereo colour transparencies. Acta Ophthal-
mol Scand 82: 656–665.
Henricsson M, Karlsson C, Ekholm L, Kai-
kkonen P, Sellman A, Steffert E & Tyrberg
M (2000): Colour slides or digital photo-
graphy in diabetes screening – a compari-
son. Acta Ophthalmol Scand 78: 164–168.
Kempen JH, O’Colmain BJ, Leske MC, Ha-
ffner SM, Klein R, Moss SE, Taylor HR &
Hamman RF (2004): The prevalence of
diabetic retinopathy among adults in the
United States. Arch Ophthalmol 122: 552–
563.
Kinyoun J, Barton F, Fisher M, Hubbard L,
Aiello L & Ferris F III (1989): Detection
of diabetic macular oedema. Ophthalmos-
copy versus photography – Early Treat-
ment Diabetic Retinopathy Study Report
No. 5. The ETDRS Research Group. Oph-
thalmology 96: 746–750.
Klein R, Klein BE & Moss SE (1984): Visual
impairment in diabetes. Ophthalmology 91:
1–9.
Knudsen LL & Skriver K (2006): A 3-dimen-
sional evaluation of the macular region:
comparing digitized and film-based media
with a clinical evaluation. Acta Ophthalmol
Scand 84: 296–300.
Martin JD & Yidegiligne HM (1998): The
cost-effectiveness of a retinal photography
screening programme for preventing dia-
betic retinopathy in the First Nations dia-
betic population in British Columbia,
Canada. Int J Circumpolar Health 57(Sup-
pl. 1): 379–382.
Massin P, Girach A, Erginay A & Gaudric A
(2006): Optical coherence tomography: a
key to the future management of patients
with diabetic macular oedema. Acta Oph-
thalmol Scand 84: 466–474.
Moller F, Hansen M & Sjolie AK (2001): Is
one 60-degree fundus photograph sufficient
for screening of proliferative diabetic reti-
nopathy? Diabetes Care 24: 2083–2085.
Neubauer AS, Welge-Lussen UC, Thiel MJ,
Alge C, Priglinger SG, Hirneiss C, Ulbig
MW & Kampik A (2003): Tele-screening for
diabetic retinopathy with the retinal thick-
ness analyser. Diabetes Care 26: 2890–2897.
Perrier M, Boucher MC, Angioi K, Gresset
JA & Olivier S (2003): Comparison of two,
three and four 45-degree image fields
obtained with the Topcon CRW6 non-
mydriatic camera for screening for diabetic
retinopathy. Can J Ophthalmol 38: 569–
574.
Sharp PF, Olson J, Strachan F et al. (2003):
The value of digital imaging in diabetic
retinopathy. Health Technol Assess 7: 1–
119.
Stefansson E, Bek T, Porta M, Larsen N,
Kristinsson JK & Agardh E (2000): Screen-
ing and prevention of diabetic blindness.
Acta Ophthalmol Scand 78: 374–385.
Stellingwerf C, Hardus PL & Hooymans JM
(2001): Two-field photography can identify
patients with vision-threatening diabetic
retinopathy: a screening approach in the
primary care setting. Diabetes Care 24:
2086–2090.
von Wendt G, Ronnholm P, Heikkila K &
Summanen P (2000): A comparison
between one- and two-field 60-degree fun-
dus photography when screening for dia-
betic retinopathy. Acta Ophthalmol Scand
78: 14–20.
von Wendt G, Heikkila K & Summanen P
(2002): Detection of retinal neovascular-
izations using 45-degree and 60-degree
photographic fields with varying 45-degree
fields simulated on a 60-degree photo-
graph. Acta Ophthalmol Scand 80: 372–
378.
WHO (2001): Definition and diagnosis of
diabetes mellitus and intermediate hyper-
glycaemia. http://www.who.org/diabetes/
publications/en (accessed August 12th,
2007)
Wilkinson CP, Ferris FL III, Klein RE et al.
(2003): Proposed international clinical dia-
betic retinopathy and diabetic macular
oedema disease severity scales. Ophthal-
mology 110: 1677–1682.
Received on March 8th, 2007.
Accepted on July 9th, 2007.
Correspondence:
Aljoscha S. Neubauer MD, MBA
Department of Ophthalmology
Ludwig-Maximilians-University Munich
Mathildenstrasse 8
80336 Munich
Germany
Tel: + 49 89 5160 5160
Fax: + 49 9131 5160 3051
Email: [email protected]
muenchen.de
Acta Ophthalmologica 2008
182