Faculty of Informatics Vienna University of Technology Institute of Computer Aided Automation PRIP Home PRIP Home
Personal tools
You are here: Home Teaching PRIP-Talks Eva Dittrich: Automatic Segmentation of Retinal Vessels and Measurement of Doppler Flow Velocity in Optical Coherence Tomography Data
Navigation
 

Eva Dittrich: Automatic Segmentation of Retinal Vessels and Measurement of Doppler Flow Velocity in Optical Coherence Tomography Data

Abschlussvortrag Diplomarbeit

What
  • Presentation
When Sep 21, 2009
from 03:15 pm to 03:35 pm
Where Sem 183/2
Add event to calendar vCal
iCal
The analysis of the vascular network in the retina is crucial for the diagnosis and study
of diseases that result in a change of the vascular pattern and the retinal blood flow,
and can even lead to blindness. Examples of diseases with these symptoms are diabetic
retinopathy, age-related macular degeneration or glaucoma, which are the three leading
causes of blindness in the USA. Optical Coherence Tomography (OCT) is a non-invasive
imaging method used for the three-dimensional in-vivo observation of the human eye’s
retinal layers. OCT allows for an observation of fine retinal vascular networks. However,
in particular the fine capillary vessels are difficult to analyze, due to noise and ambiguous
appearance in the data.
The aim of this work is to detect retinal vessels in these OCT data. Based on an
initial coarse vessel detection, a set of candidate points is retrieved, and a probabilistic
kernel reflects the mutual relations of these vessel candidate points. Due to the noise
in the OCT data it is necessary to integrate the information of larger vessel segments
in the detection process to achieve a robust segmentation. This is done by embedding
the initial candidate positions in a diffusion map that captures the local structure and
mutual spatial relations of the vessel points. The positions in this map can be used to
efficiently distinguish between vessels and background noise. The second contribution of
this thesis is a method to measure the blood flow velocity inside the detected vessels by
including Doppler information that is gained during the acquisition of the OCT data.
Since this given velocity information is only captured along the signal beam, and thus
does not reveal the actual blood flow speed inside the vessels, a correction has to be
performed according to the detected centerlines of the vessels.
Experimental results show a substantial improvement compared to existing vessel detec-
tion methods. In addition, the resulting blood flow speed is a more accurate estimation
of the blood support and circulation inside the vessels than the measured speed without
knowledge about vessel direction; thus, it enables an investigation of the actual blood
flow speed inside the retinal vessels.
Document Actions