Object Segmentation in Speckle Noise

A novel filtering approach drastically reduces speckle noise in OCT.

October 2010 - August 2011
Cyrill Gyger

Speckle noise is a massive distortion in Optical Coherence Tomography. This characteristic noise, caused by optical interferences, disturbs the visual image quality and complicates the computational post processing, like image segmentation. In this master thesis the reduction of speckle noise in 3D OCT images and the subsequent image segmentation was prospected. A novel filtering approach was developed which is able to reduce speckle noise in OCT images drastically. Based on this algorithm a specialized processing chain was designed that successfully realizes the segmentation of choroidal vessel structures in high-resolution 3D OCT volumes.