Aktuelle Tagungsinformationen
News and Updates
Anmeldung zur Tagung
Registration
Hotelbuchung
Hotel Registration
Grußwort
Welcome address
Beteiligte Gesellschaften
Societies involved
Eröffnung des Kongresses
Opening Ceremony
Preise
Awards
Wissenschaftliches Programm
Scientific program
Posterpräsentationen
Poster Presentation
Kurse
Courses
Begleitende Veranstaltungen
Collateral Events
Rahmenprogramm
Social program
Jubiläumsparty
Jubilee Party
DOG Information
DOG Information
Allgemeine Informationen
General Information
Autorenindex
Index of Authors
Ausstellerliste
Exhibitors
Sponsoren
Sponsors
Teilnahmegebühren
Registration fees
Impressum
DOG Homepage
Simulation-based Analysis on Automated Classification of HRT Images
1Adler W., 2Hothorn T., 2Lausen B., 3Paulus D.,
1Friedrich-Alexander-Universität, Institut für Medizininformatik, Biometrie und, Epidemiologie (IMBE) (Erlangen)
2Friedrich-Alexander-Universität, Institut für Medizininformatik, Biometrie und Epidemiologie (IMBE) (Erlangen)
3Institut für Computervisualistik, Universität Koblenz-Landau (Koblenz)
Purpose: Evaluate the method by Swindale et al. (2000) for automated classification of images from the Heidelberg Retina Tomograph (HRT) based on simulated images.
Method: As a precondition to medical analysis of HRT-images, the optic nerve head has to be outlined manually in clinical practise. Swindale et al. (2000) suggest automated classification of HRT images by a nonlinear approximation of the image function using the fitted parameters of a nonlinear model. To evaluate the ability of this method to classify glaucomatous and normal eyes, we use a simulation model, which allows to create topography images of the optic nerve head of different forms (normal, glaucomatous and various progressive forms). We compare linear discriminant analysis, classification trees, bagging (Breiman, 1996), and double-bagging (Hothorn & Lausen, 2002). The methods are illustrated by data of a case control study (Mardin et al., 2002).
Results: Bagging decreases error rates for glaucoma classification using the parameters provided by the method of Swindale.
Conclusions: The results are of major importance for the development of physician independent screening methods.
References: Breiman L (1996): Bagging Predictors. Machine Learning, 24(2), 123-140. Hothorn T, Lausen B (2002): Bagging combined classifiers. In: Proceedings of the 8th Conference of the International Federation of Classification Societies, July 16-19, 2002, Cracow, Poland, Springer, Heidelberg. Mardin CY, Hothorn T, Peters A, Jünemann AG, Michelson G, Lausen B (2002) : New Glaucoma classification method based on standard HRT parameters by bagging classification trees, submitted. Swindale NV, Stjepanovic G, Chin A and Mikelberg FS (2000): Automated analysis of normal and glaucomatous optic nerve head topography images. Investigative Ophthalmology and Visual Science, 41(7), 1730-1742.