The main goal of this research project is to develop automatic methods for early detection of various eye diseases, such as glaucoma, diabetic retinopathy or papilla oedema.
For the evaluation of these methods the extensive Erlanger Glaucoma Database of the Collaborative Research Center 539 can be accessed. In addition to patient medical records, anamnesic and diagnostic data the database also stores color fundus photographs and topography images of the retina obtained by confocal laser scanning.

In a previous research project a multimodal system for segmenting the optic nerve head was developed at the Chair for Pattern Recognition. This approach allows to measure regions of the eye ground in order to diagnose glaucoma. However, this segmentation based method can and should be improved.

Currently, we assess the feasability of using appearance based methods, such as principal component analysis (PCA) or linear discriminant analysis (LDA), to extract characteristic features from the images.

My current research projects are:

• Automated Glaucoma Risk Index
• Papilla Change Detection

Examples of retinal images: digital fundus photographs field of view 45° and 20°,

Heidelberg Retina Tomograph reflectance and topography images (left to right).