Department InformatikTechnical Reports / ISSN 2191-5008

Frank Lauterwald, Christoph P. Neumann, Richard Lenz, Anselm G.Jünemann, Christian Y. Mardin, Klaus Meyer-Wegener, Folkert K.Horn

The Erlangen Glaucoma Registry: a ScientificDatabase for Longitudinal Analysis of Glaucoma

Technical Report CS-2011-02

December 2011

Please cite as:

Frank Lauterwald, Christoph P. Neumann, Richard Lenz, Anselm G. Jünemann, Christian Y. Mardin, Klaus

Meyer-Wegener, Folkert K. Horn, “The Erlangen Glaucoma Registry: a Scientific Database for Longitudinal Analysis of

Glaucoma,” University of Erlangen, Dept. of Computer Science, Technical Reports, CS-2011-02, December 2011.

Friedrich-Alexander-Universität Erlangen-NürnbergDepartment Informatik

Martensstr. 3 · 91058 Erlangen · Germany

www.informatik.uni-erlangen.de

The Erlangen Glaucoma Registry: a ScientificDatabase for Longitudinal Analysis of Glaucoma

Frank Lauterwald1, Christoph P. Neumann1, Richard Lenz1, Anselm G. Jünemann2,Christian Y. Mardin2, Klaus Meyer-Wegener1, Folkert K. Horn2

Data Management1Dept. of Computer Science, University of Erlangen, Germany

2Dept. of Ophthalmology, University of Erlangen, Germanyfrank.lauterwald@informatik.uni-erlangen.de

Abstract—Background: Due to the slow progression ofthe Glaucoma disease, a large study population and long-time observations are needed to gain insights into itslong-term effects and progression rates. Since modalitiescan export data in machine-readable formats, statisticalanalyses of the large number of examinations is feasible.These data have been integrated in a central patientregistry, the Erlangen Glaucoma Registry (EGR).

Objectives: The primary focus of the EGR systemdesign has been its fitness for analyses. It holds almostall the available research data for registered glaucomapatients. This allows for cross-sectional and longitudinalobservations and for evaluation of prognostic validity ofdiagnostic procedures.

Methods: An adequate technology for integration ofdata and flexibility in data analysis is a database man-agement system (DBMS). Here, a careful schema designis mandatory. Adding new modalities leads to schemamodifications which are supported by defining a coredatabase schema and attaching all data to this core.On that basis, a large number of modalities have beenconnected to the EGR.

Results: The registry contains data of 1,400 patients inthe main longitudinal study. It has successfully helped inscientific research, as can be seen in a large number ofpublished papers. For example, validation of new sensoryphysiological methods requires patients with reliable diag-noses. The existence of a well-documented patient collectivefacilitated finding such patients. Conclusions: The EGRholds a unique amount of available data gathered in largelongitudinal studies. It was successful in terms of medicalresults obtained. Developed as an evolutionary system, itcan easily be extended.

Index Terms—Glaucoma, Medical Record Systems(computerized), Registries, Longitudinal Studies

I. INTRODUCTION

Glaucoma is a slowly progressing eye disease thatis accountable for more than one thousand new cases

Fig. 1. Progression Charts

of blindness in 2003 in Germany [1] and for 120,000cases of blindness in the U.S.A., which is 9% to 12% ofall cases of blindness [2]. Due to the slow progressionof the disease, a large study population and long-timeobservations are needed to gain insights into its long-term effects and progression rates [3]. Fig. 1 shows anexample of an individual glaucoma progression chart.It contains the history of several measurements for onepatient and allows for a quick overview. New methodsfor early detection can only be evaluated retrospectively,as there might be no sufficiently well established methodavailable to compare the new methods against. This,too, requires long-term observations to correlate laterprogression with findings obtained several years earlier.

During the nineties, more and more modalities allowedto export data in machine-readable formats. This madestatistical analyses of the large number of examinations alot more feasible. It was expected that the emergence ofhistorical examination data would gradually allow for de-ductive revisal of the diagnostic methods that have beenapplied to patients some years ago. For example, patientswith actual glaucoma that remained unrecognized duringearly detection screening are selected for a re-audit. Thisallows to evaluate whether current diagnostic methodswould have provided correct indication of the glaucomausing the historical modalities’ data.

Information technology has fostered medical co-operation by providing technical means for the cor-relation of data from different laboratories [4]. Suchcollaboration has naturally led to the creation of a centralpatient registry, the Erlangen Glaucoma Registry (EGR).The EGR as a central platform for glaucoma diagnosticshas been initiated by Prof. Naumann in 1991 and isremarkable for both its patient and research scale. Itfacilitates long-term research and has enabled a numberof new analyses, e.g. [5].

A few similar approaches can be located in the liter-ature and in the World-wide Web. However, a registrywith the focus on Glaucoma could hardly be found.An early article [6] describes a glaucoma database withmanual input. The technology of that time was limitedcompared to that of today, and no direct input frommodalities was possible. There is no description of pos-sible longitudinal studies. The Web page of the NationalEye Database (NED) of the Association of Clinical Reg-istries in Malaysia (ACRM) [7] mentions a "GlaucomaRegistry (in progress)", but no further information isavailable. The Australian and New Zealand Registry ofAdvanced Glaucoma [8] has collected DNA samples of400 patients so far to identify and disseminate clinicaland epidemiological risk factors. The EUREQUO projectbuilds a registry for cataract and refractive surgery [9].

For some diseases, registries are very common. Proba-bly the most well-known are cancer registries [10] whichdate back to the 1920s and can be either epidemiologicalor clinical. In contrast, all glaucoma registries that weare aware of are clinical registries. Registries in generalfor other diseases such as the ICU incident registry [11]usually concentrate on input from various kinds of usersand thus emphasize user-friendly interfaces. EuCliD [12]is a tool for supervising selected quality indicators ofabout 200 European dialysis centers. Its data model israther simple and consists of just 5 modules. Thus, ituses a Lotus Notes-based flat-file database instead of

a relational database. The number of patients equals14,000. Data can be entered manually or by interfaces toexisting local data management systems. Data analysisis performed with statistical tools like SPSS.

Some registries are designed specifically for clinicaltrial design. Peelen and coworker [13] showed how suchregistries can foster research by choosing good eligibilitycriteria for balancing statistical power and study costs.The EGR has also been used for recruiting participantsfor individual studies, but this is only a side-effect ofEGR.

In contrast to most other approaches, we have con-centrated on the import of data from different modalitiesand lab applications, thus enabling automatic data entrywherever possible. The objectives of this paper are topresent the EGR’s architecture and data model and todescribe its contribution in achieving a-posteriori revisalof diagnostic methods. Furthermore, the additional ben-efits of an integrated IT system in terms of automationof tasks and the recruitment of patients for new studiesare described.

II. OBJECTIVES

The design of the EGR system had several objectivesin mind. Primary focus has been its fitness for analyses.By integrating more and more data from modalities andlab applications into this system, it has been enriched tohold any available research data for registered glaucomapatients. That allows for cross-sectional and longitudinalobservation of patients with open angle glaucoma orglaucoma suspect. The rationale has always been theevaluation of diagnostic and prognostic validity of mor-phometrical, sensorical and hemodynamical diagnosticprocedures (e.g. [14]). Beyond determining correlationsbetween different diagnostic methods, this registry fur-thermore allows to find patients that meet the eligibilitycriteria for new studies. Finally, the EGR as central inte-gration platform also aims to reduce workload for users,e.g. with automated discharge letters. For the success ofthe EGR, usability is crucial, because it is integratedin the daily work of medical end-users who do nothave deep technical expertise. New examination typesemerge quickly, so the EGR must be a responsive ITinfrastructure that can react adequately to growing needs.This requires the capability to add new functionality withminimum effort and without the risk of breaking thedesign.

III. METHODS

Standardized inspection of fundus photographs ac-cording to Jonas [15], tonometry and white-on-whiteperimetry form the goldstandard of the EGR. Theirdata had to be included from the beginning togetherwith anamnesis data and clinical findings as well asmorphological data and results from sensory tests thatwere developed at the University Eye Hospital in Er-langen, e.g. temporal contrast sensitivity and blue-on-yellow Visual Evoked Potentials (VEP). Later, Heidel-berg Retina Tomography (HRT), Scanning Laser DopplerFlowmetry (SLDF), Nerve Fiber Analyzer (GDx), Opti-cal Coherence Tomography (OCT), Doppler Sonography,Electro-Retinograms (ERG), and Frequency-DoublingTechnology (FDT) were added. For a separate group ofpatients, fundus images from the Kowa camera, HRTand FDT data have been collected. In addition to thesedata, anamnestic data and diagnostic findings are alsoincluded. For many patients, the information has beengathered for an extensive period. Evaluations are sup-ported by mechanisms to select, combine, and extractdata in a format for further analysis (i.e. SPSS). Thisallows, for example, evaluation of functional-structuralrelationship [14], a comparison of central corneal thick-ness and intra-ocular pressure (IOP) in patients withlarge optic discs and physiological macrocup [16], andthe correlation of optic disc morphometry with confocalSLDF measurements [17]. An adequate technology forthe integration of data from many sources and forflexibility in data analysis is a database managementsystem (DBMS). It provides data independence, multi-user operation, data integrity, and access control. Re-lational DBMS are well-established and suffice for therequirements of the EGR. However, a careful schemadesign is mandatory. It creates data structures appropriatefor easy integration of input data and neutral to thedifferent evaluations to come. Adding new modalitiesor data input from other sources often leads to schemamodifications which are a tough problem in the appli-cation of a DBMS. However, they can be supported bydefining a core database schema and attaching all datato this core (Fig. 2). The core schema is intentionallykept simple and mostly consists of tables for patientsand inspections. Each data element is either patient- orinspection-specific. There may be an arbitrary numberof patient-specific data elements per patients. Similarly,any number of inspection-specific data elements may bestored and a patient may be examined any number oftimes. The simplicity of the core allows it to remain

Fig. 2. Core database schema

unchanged in spite of the continuous evolution of thesystem. All schema modifications can then be done bysimply adding tables. Stable core data structures areof paramount importance allowing meaningful analysescovering long periods of time. The complete schemacontains about 70 tables. For known analyses that useEGR data, views have been defined that combine allrelevant data in one table. These are complex queriesstored in the database and given a name, so they appearas a stored table. This significantly eases the use of thedata while at the same time allowing for changes in thebase tables. For completely new analyses, experiencedusers may use the query language - typically SQL - totailor the extracted data set to their particular needs.

To determine the satisfaction of users with the sys-tem and identify areas for further improvement, weconducted a survey by interviewing physicians and re-searchers with the help of a standardized questionnaire.The feedback from the survey is discussed in Sect. V-A.The basic architecture of the EGR can be verticallydivided into four layers (Fig. 3). The top layer consistsof the data sources. These can be subdivided into threegroups: modalities, current applications and legacy ap-plications (with their own local databases). Data frommodalities and legacy applications is filtered through thesecond layer, while current applications already containthe checks and refinement that are performed there.The second layer-called the integration layer-comprisesa classical ETL staging area [18]. Its duty is to performdata cleansing, quality checks and schema transformationso that the data matches the schema of the EGR. Individ-

Fig. 3. Architecture of the Erlangen Glaucoma Registry

ual software needs to be written for each modality andlegacy application. This is necessary because there are nostandardized data formats for ophthalmologic modalities.Instead, all vendors define their own export format (e.g.CSV, XML or Excel) that may change with new versionsof a modality or even with software updates. The thirdlayer is the central database. This layer is where the datais actually stored. Finally, the fourth layer is the one theEGR is all about. Its purpose is to facilitate research byhelping in analyzing the data. It allows the export of EGRdata into various other formats and applications. We seethe views targeted at analyzing data as part of this layer.The benefit of such architecture is flexibility: In each ofthe layers modifications are possible that do not affectthe other layers. Data is entered directly into the systemby three physicians and three technical assistants, witha number of other people entering data into modalitiesand legacy databases. These data are then integrated intothe central database via automated batches.

IV. RESULTS

This section is divided into three subsections. Sect.IV-A describes how different data sources were inte-

grated as well as the amount of data contained in theEGR. Sect. IV-B focuses on medical results obtainedwith the help of the EGR, while Sect. IV-C describesimprovements in processes.

A. Data Integration

The integration of data from various heterogeneoussources has been successful and created a central dataresource for numerous evaluations. The combinationof diagnostic findings with the input data from IOP,papillometry and perimetry form the goldstandard ofglaucoma diagnosis. The availability of these data in theEGR allows to understand the physician’s decision. Fur-thermore, the goldstandard data can be used to measurethe diagnostic and prognostic validity of other ophthal-mological diagnostic procedures. The registry currentlycontains data about 1,400 patients and study participantsin the main longitudinal study. Among them are 365normals (i.e. healthy subjects) that serve as controlgroup. The patients usually receive follow-up inspectionsevery year, while healthy participants are examined lessfrequently. The total number of inspections in this study

exceeds 6,400, the average number of inspections perpatient is about 4.4.

This number is lower than what might be initiallyexpected, but can be easily explained by two factors:First, patients drop out of the study for various reasons,one of which is age. The median age of patients is about61.5 years. Thus, travelling to the research center for twodays of inspections may become too tedious for somepeople. Second, new subjects continue to be recruited forthe study. Newly recruited subjects simply cannot havereceived a large number of inspections yet. For example,256 patients have received 10 or more examinations,with a maximum of 19 examinations for some patients.The patients are usually reexamined on a yearly basis.Two large screening studies account for another 3,181subjects. Additionally, there are more than 4,000 patientsfrom several individual studies.

B. Implications for medical research

The Erlangen Glaucoma Registry has successfullyhelped in scientific research. This can be seen in a largenumber of published papers referring to the EGR (e.g.[14], [17], [19], [16]). For example, several new sensoryphysiological methods have been developed. Validationof these methods requires patients with reliable di-agnoses. The existence of a well-documented patientcollective facilitated finding such patients ([20], [21],[22], [23], [24]). Several co-operations with industrialpartners led to the introduction and improvement ofnew modalities, e.g. HRT (1995), GDX (1997) or OCTSpectralis (2008). Again, the EGR was used to validatemodalities’ results and to define norm values for theirmeasurements ([5], [17], [25]). The long-term nature ofthe EGR made it possible to test methods for short-and long-term reproducibility ([26]). Additionally, itallowed a posteriori-analyses of different therapies ([27])as well as the influence of physiological or lifestyle-related factors [16], [28]. Finally, it became possible totest different methods for their predictive value ([17],[19], [29], [30], [31], [32]). This, too, requires observinga large patient collective over a long time, as progressionrates are relatively low.

C. Effects on User Workload

Besides allowing scientific analyses, another goal ofthe EGR is to assist its users in their daily work. Thisled to the development of several custom applications:Making and tracking appointments is facilitated by theability to list all patients that should be re-examinedin the near future. This helps ensuring that glaucoma

patients are re-examined on a yearly basis. Dischargeletters containing all relevant examination data as wellas a therapy recommendation can be automatically gen-erated. This is easily possible as all needed data isalready stored in the database. This significantly reducesthe effort of writing discharge letters. These letterscontain the diagnosis and recommended therapy andare sent to a physician in private practice who treatsthe patient between examinations at the eye-hospital.Individual Glaucoma Progression Charts (Fig. 1) allowthe physician to quickly gain insights on a patient’smedical history. They are also used as an educational toolin weekly internal training sessions as the progressioncharts allow to easily present a certain case along withits special characteristics.

V. DISCUSSION

While there is always potential for enhancements, theEGR has worked reliably for several years. The primarypurpose of EGR as always been driven by the specificmedical goals. As a centralized system, the EGR fosterscontrol over the data. The use of a DBMS necessitatedconsolidation efforts between the participants and acomputer scientist is required to keep the system upand running. Yet, any solution with local data copiesscattered in a desktop environment, or - even worse -a file-based data organization, suffers from redundantdata, conflicts between users, and erroneous data withconsistency violations. The consolidated data schemaimproves data quality by preventing inconsistent data anduncontrolled data redundancy. Especially in medical andresearch environments with a high turnover rate, only acentral system allows to find and interpret older data thatmight otherwise get lost after first publication or whenscientists have long moved on to different places. Thisdirectly helps to achieve the main goal of the EGR -namely fostering medical research. Data quality andinterpretability are essential for achieving meaningfulresults - especially for long-term studies. Additionally,a DBMS-based system greatly facilitates new analysesthrough its powerful and well-understood query capabil-ities. Especially combining data from multiple sources isgreatly facilitated compared to the other means of datastorage that were mentioned above. The successful uti-lization of the EGR for medical research shows that thegoal of being a reliable research tool has been reached.While primarily designed as an integration platform tofacilitate data analysis, it has proven to be an efficienttool for many other purposes, like the recruitment of

patients, deriving norm values for modalities, testingpredictive values, and its educational exertion.

A. User Feedback

In order to determine users’ satisfaction with thesystem, we conducted an informal survey and askedthem about the accomplishments and shortcomings ofthe EGR. This survey indicated the following: In general,the users seem to be quite satisfied with the system andbug numbers are perceived to be low. When specificallyasked about the shortcomings of the system, one issuestood out remarkably: There is a need to explore theexisting data more easily. This includes e.g. getting aquick overview of parts of the existing data, writingtheir own queries or determining the number of patientseligible for a planned study. The necessity to ask an ITexpert for a filtered or aggregated list of patients andexaminations and post processing in a statistics pack-age is seen as too cumbersome for some cases. Whilethis process is perceived as adequate when analyzingresults and preparing publications, there exists a desireto quickly do some preliminary checks and aggregationsin order to identify interesting patterns and promisingquestions for further analysis. It can be concluded frommedical researchers’ responses, that the data evaluationsthey want to perform are highly individual and requiregreat flexibility.

B. System Outlook

The survey has given valuable hints for further devel-opment. The shortcomings in ad-hoc support are natural:such analyses require precise communication betweenthe system administrator and the medical experts inorder to gain an accurate definition of the emergentrequirements. Another factor is that such analyses areexecuted only when conducting a study but not in dailyroutine, thus, they are only sporadically executed andhave less chance to mature.

It seems impractical to build both powerful and easy-to-use query facilities into the system. Yet, in order toprovide adequate support, serviceable database viewswill be identified and created that mitigate the com-plexity of querying a complex database schema withseveral indirections between relevant tables. It also seemspromising to foster trainings for medical researchers inthe application of existing graphical query tools like MS-Access. In addition, we plan to allow for parameterizedanalyses, such that users can customize the most fre-quently utilized querys.

VI. CONCLUSION

In this paper, we described the architecture and currentstate of the Erlangen Glaucoma Registry. It serves as acentral repository for all data gathered in large longitu-dinal studies. We have shown the success of the EGR interms of the unique amount of available data and medicalresults obtained with its help. The EGR was alwaysplanned as an evolutionary system that could easily beextended over a long period of time. While primarilydesigned as an integration platform to facilitate dataanalyses, it has proven to be an efficient tool for manysecondary purposes as well. Ideas for further analysisand improvement of the system have been presented.Last but not least, a centralized yet flexible systemhelps keeping control over the data. Despite the highinitial implementation costs, the project has proven to besuccessful. We see the EGR as a solid basis for futureanalysis of glaucoma.

VII. ACKNOWLEDGEMENTS

Thanks to Prof. Naumann and Prof. Martus for startingthe EGR project.

REFERENCES

[1] C. Knauer and N. Pfeiffer, “Erblindung in Deutschland –heute und 2030,” Der Ophthalmologe, vol. 103, no. 9, pp.735–741, 2006. [Online]. Available: http://www.springerlink.com/content/144873121q037502/

[2] Glaucoma Research Foundation. Glaucoma Facts andStats. Cited 2011 Dec 13, San Francisco. [Online].Available: http://www.glaucoma.org/glaucoma/facts-statistics/glaucoma-facts-and-stats.php

[3] L. Horbach, Erlanger Reihe der Medizinischen Statistik undInformationsverarbeitung. Palm und Enke, Erlangen, 1996,ch. Biometrisch orientierter Forschungsansatz und Planung derErlanger klinischen Studie über chronische Glaukome.

[4] L. Horbach and M. Wisse, Erlanger Reihe der MedizinischenStatistik und Informationsverarbeitung. Palm und Enke, Er-langen, 1996, ch. Technik der Datenverarbeitung.

[5] W. Adler, A. Peters, and B. Lausen, “Comparison of classifiersapplied to confocal scanning laser ophthalmoscopy data,” Meth-ods of Information in Medicine, vol. 47(1), pp. 38–46, 2008.

[6] R. H. Greenfield, M. A. Kass, and J. P. Livingston, “A Com-puterized Glaucoma Data Base,” Archives of Ophtalmology, vol.95(8), pp. 1365–1367, 1977.

[7] National Eye Database (NED). Association of ClinicalRegistries, Malaysia. Updated 2007 Dec 3, cited 2011Dec 13. [Online]. Available: http://www.acrm.org.my/affiliatedDatabases_eyeHealth%20.htm

[8] The Australian and New Zealand Registry of AdvancedGlaucoma. The Eye Foundation. Surry Hills NSW 2010Australia. Updated 2011 Nov 17, cited 2011 Dec 13.[Online]. Available: http://www.eyefoundation.org.au/projects/research-projects/90-registry-of-advanced-glaucoma

[9] European Registry of Quality Outcomes for Cataract and Re-fractive Surgery. EUREQUO Project. http://www.eurequo.org/.Cited 2011 Dec 13.

[10] World Health Organization, International Agency for Researchon Cancer, World cancer report. Lyon: IARC Press, 2003.

[11] S. van der Veer, R. Cornet, and E. de Jonge, “Design and im-plementation of an ICU incident registry,” International Journalof Medical Informatics, vol. 76(2-3), pp. 103–108, 2007.

[12] H. Steil, C. Amato, C. Carioni, J. Kirchgessner, D. Marcelli,A. Mitteregger, V. Moscardo, G. Orlandini, and E. Gatti,“EuCliD–a medical registry,” Methods of Information inMedicine, vol. 43(1), pp. 83–88, 2004.

[13] L. Peelen, N. Peek, E. de Jonge, G. Scheffer, and de Keizer N.F.,“The use of a registry database in clinical trial design: assessingthe influence of entry criteria on statistical power and number ofeligible patients,” International Journal of Medical Informatics,vol. 76(2-3), pp. 176–183, 2007.

[14] C. Mardin, A. Peters, F. Horn, A. Jünemann, and B. Lausen,“Improving glaucoma diagnosis by the combination of perime-try and HRT measurements,” Journal of Glaucoma, vol. 15(4),pp. 299–305, 2006.

[15] J. Jonas, G. Gusek, and G. Naumann, “Optic disc, cup andneuroretinal rim size, configuration and correlations in normaleyes,” Investigative Ophtalmology & Visual Science, vol. 29(7),pp. 1151–1158, 1988.

[16] A. Viestenz, N. Wakili, A. Jünemann, F. Horn, and C. Mardin,“Comparison between central corneal thickness and IOPin patients with macrodiscs with physiologic macrocupand normal-sized vital discs,” Graefes Arch. Clin. Exp.Ophthalmol., vol. 241, pp. 652–655, Aug 2003. [Online].Available: http://www.ncbi.nlm.nih.gov/pubmed/12898281

[17] J. Jonas, J. Harazny, W. Budde, C. Mardin, K. Papastathopoulos,and G. Michelson, “Optic disc morphometry correlated withconfocal laser scanning Doppler flowmetry measurements innormal-pressure glaucoma,” Journal of Glaucoma, vol. 12, pp.260–265, Jun 2003. [Online]. Available: http://www.ncbi.nlm.nih.gov/pubmed/12782846

[18] P. Vassiliadis, A. Simitsis, and S. Skiadopoulos, “Conceptualmodeling for ETL processes,” in DOLAP ’02: Proceedings ofthe 5th ACM international workshop on Data Warehousing andOLAP. New York, NY, USA: ACM, 2002, pp. 14–21.

[19] P. Martus, A. Stroux, W. Budde, C. Mardin, M. Korth, andJ. Jonas, “Predictive factors for progressive optic nerve damagein various types of chronic open-angle glaucoma,” AmericanJournal of Ophtalmology, vol. 139(6), pp. 999–1009, 2005.

[20] M. Korth, A. Jünemann, F. Horn, A. Bergua, C. Cursiefen,I. Velten, W. Budde, M. Wisse, and P. Martus, “Synopsis ofvarious electrophysiological tests in early glaucoma diagnosis–temporal and spatiotemporal contrast sensitivity, light- andcolor-contrast pattern-reversal electroretinogram, blue-yellowVEP,” Klinische Monatsblätter für Augenheilkunde, vol. 216(6),pp. 360–368, 2000.

[21] A. Stroux, P. Martus, W. Budde, F. Horn, A. Jünemann,M. Korth, and J. Jonas, “Sequential classification in glaucomadiagnosis,” Graefes Arch. Clin. Exp. Ophthalmol., vol. 241(4),pp. 277–283, 2003.

[22] B. Link, A. Jünemann, R. Rix, O. Sembritzki, A. Brenning,M. Korth, and F. Horn, “Pupillographic Measurements withPattern Stimulation: The PupilŠs Response in Normal Subjectsand First Measurements in Glaucoma Patients,” InvestigativeOphtalmology & Visual Science, vol. 47, pp. 4947–4955, 2006.

[23] N. Wakili, F. K. Horn, A. G. Jünemann, N. X. Nguyen, C. Y.Mardin, M. Korth, and J. Kremers, “The photopic negativeresponse of the blue-on-yellow flash-electroretinogram in glau-comas and normal subjects,” Documenta Ophtalmologica, vol.117(2), pp. 147–154, 2008.

[24] T. Lindenberg, A. Peters, F. Horn, B. Lausen, and M. Korth,“Diagnostic value of multifocal VEP using cross-validation andnoise reduction in glaucoma research,” Graefes Arch. Clin. Exp.Ophthalmol., vol. 242(5), pp. 361–367, 2004.

[25] W. Schrems, C. Mardin, F. Horn, A. Jünemann, and R. Laem-mer, “Comparison of Scanning Laser Polarimetry and OpticalCoherence Tomography in Quantitative Retinal Nerve FiberAssessment,” Journal of Glaucoma, vol. 19(2), pp. 83–94, 2010.

[26] F. Horn, B. Link, C. Mardin, A. Jünemann, and P. Martus,“Long-term reproducibility of screening for glaucoma withFDT-perimetry,” Journal of Glaucoma, vol. 16, pp. 448–455,Aug 2007. [Online]. Available: http://www.ncbi.nlm.nih.gov/pubmed/17700287

[27] F. Michalk, G. Michelson, J. Harazny, U. Werner, W. Daniel,and D. Werner, “Single-dose nimodipine normalizes impairedretinal circulation in normal tension glaucoma,” Journal ofGlaucoma, vol. 13(2), pp. 158–162, 2004.

[28] C. Cursiefen, M. Wisse, S. Cursiefen, A. Jünemann, P. Martus,and M. Korth, “Migraine and Tension Headache in High-Pressure and Normal-Pressure Glaucoma,” American Journalof Ophtalmology, vol. 129(1), pp. 102–104, 2000.

[29] R. Laemmer, S. Schroeder, P. Martus, A. Viestenz, andC. Mardin, “Quantification of neuroretinal rim loss usingdigital planimetry in long-term follow-up of normals andpatients with ocular hypertension,” Journal of Glaucoma,vol. 16, pp. 430–436, Aug 2007. [Online]. Available:http://www.ncbi.nlm.nih.gov/pubmed/17700284

[30] F. Horn, J. Jonas, W. Budde, A. Jünemann, C. Mardin, andM. Korth, “Monitoring Glaucoma Progression with VisualEvoked Potentials of the Blue-Sensitive Pathway,” InvestigativeOphtalmology & Visual Science, vol. 43, pp. 1828–1834, 2002.

[31] F. Horn, N. Nguyen, C. Mardin, and A. Jünemann, “Combineduse of FDT-perimetry and polarimetric measurements of retinalnerve fiber layer in glaucoma detection,” American Journal ofOphtalmology, vol. 135(2), pp. 160–168, 2003.

[32] F. Horn, A. Brenning, A. Jünemann, and B. Lausen, “Glau-coma Detection with Frequency Doubling Perimetry and Short-wavelength Perimetry,” Journal of Glaucoma, vol. 16(4), pp.363–371, 2007.