Hotelbuchung
Hotel Registration
Grußwort
Welcome address
Beteiligte Gesellschaften
Societies involved
DOG Information
DOG Information
Eröffnung des Kongresses
Opening Ceremony
Preise
Awards
Ablauf der Tagung 2003
General overview of congress
Lageplan der Räumlichkeiten
Map of Congress Center
Wissenschaftliche Themen
Scientific topics
Symposien
Symposia
Wissenschaftliches Programm
Scientific program
Posterpräsentationen
Poster Presentation
Kurse
Courses
Begleitende Veranstaltungen
Accompanying program
Arbeitssitzungen
Working sessions
Rahmenprogramm
Social program
Allgemeine Informationen
General Information
Autorenindex
Index of Authors
Industrieaussteller
Commercial exhibitors
Sponsoren
Sponsors
Impressum
DOG Homepage
Abstract
Abstract
Software-based microscope tracking for anterior segment eye surgery
Lischka T.1, Schulz S.2, Grigat R.-R.2, Richard G.1
1University Eye Hospital Hamburg-Eppendorf, Hamburg; 2Technical University Hamburg-Harburg, Hamburg
Purpose: The use of an operating microscope in surgical interventions requires frequent readjusting to achieve an optimal image of the surgical field. Software-based microscope tracking may relieve surgeons from this problem and increase the number of adequate images.
Method: The circular iris was chosen as a constant feature for image centering. Using Hough Transformation the midpoint of the iris in the microscopic field was computed continuously within short intervals (< 1 sec.). The data were compared with threshold values of quality parameters. The automatically computed midpoints were compared with manually measured iris midpoints from previously saved original images.
Results: 3.483 (97.9%) of the computed iris midpoints were correct, 54 (1.5%) wrong. After optimal adjusting of the threshold values, 98.1% of the correctly computed iris midpoints were accepted, 100% of the wrong computed values were rejected. The mean deviation of the computed iris midpoints from the manually determined iris midpoints was 2.8 pixel for the correctly computed values versus 62.5 pixel for the wrong computed values.
Conclusions: The used image processing software proved to be robust and reliable under realistic conditions. The computed iris midpoints can be used for automatic tracking