Erweiterung eines Gefäßmodells zur Validierung von Gefäßmessalgorithmen in der Retina sowie theoretische und experimentelle Untersuchungen zu Messfehlerquellen der Gefäßanalyse

Content Provider	Semantic Scholar
Author	Saleh, Kutaiba
Copyright Year	2009
Abstract	AbstractAs a „diagnostic window“ the retinal vessel system provides an insight into the human body; it not only delivers diagnostic evidence for vascular eye diseases, but it also allows conclusions reg. systemic vascular diseases. Thereby retinal vessel analysis records vascular changes esp. by measuring retinal vessel diameters and their dynamic response to different measurement and stimulation conditions. Diverse measurement methods are implemented for retinal vessel diameter measurements.In the application of those measurement methods a multitude of measurement parameters (e. g. the optical properties of the measurement chain, illumination parameters…) as well as biological influencing variables (spatiotemporal biological variability, head and eye movements) are determining the measurement conditions and for this reason are sources of measurement errors.Objective criteria for the verification of measurement results as well as for comparing different measurement methods can be developed by means of vessel imaging models.Since physical models – esp. with respect to their complexity and variability – are not reasonably realisable with sufficient accuracy and adequate practicability, in this research work a numerical model has been developed. By use of this model the impact of a multitude of technical and biological influencing variables on the measurement results can be simulated with high flexibility.In the context of this work an existing numerical basic model has been comprehensively restructured, upgraded, considerably expanded and implemented. Thereby the mapping procedure of the retinal vessels up to the recorded pixel image is simulated close to reality. The model expansion allows for the consideration of additional influencing parameters. Interference in the vessel representation as well as the optical properties of blood (venous, arterial) are taken into consideration in the process. Furthermore additional parameters for specifying the parameters of imaging sensors are implemented. Subsequently an output alternative for the simulated vessel images into an image file has been provided. This opens up the opportunity to apply any measurement methods to these (simulated) images and thus to compare the results of vessel diameter measurements.By applying this model now an evaluation of measurement algorithms for vessel diameter investigation becomes possible. Thereby now also exists the chance for separately acquiring the influence of different model parameters on the linearity of the measurement relationship. For this purpose different model parameters in real fundus images are analysed for experimentally determining their range of values. Further domains are drawn from the literature. Within these domains the differences between measured values and reference values are recorded and with it the methodical error and a correction function for the in the model integrated automatic measurement method is determined. Additionally, the influence of variations of selected model parameters (e. g. contrast, brightness, optical resolution) on the vessel diameter is demonstrated. The influence of stochastic variations of model parameters can be quantified and systematically simulated by the model. Furthermore, by freely adjustable parameters the model makes the determination of an operational range possible in which the measurement method operates reliably. Thus quality standards for eye fundus imaging can be derived which for a measurement method to be tested should be necessarily met to guarantee the required accuracy and reliability of the measured data.
File Format	PDF HTM / HTML
Alternate Webpage(s)	https://www.db-thueringen.de/servlets/MCRFileNodeServlet/dbt_derivate_00016848/ilm1-2009000045.pdf
Alternate Webpage(s)	http://www.gbv.de/dms/ilmenau/toc/599083557.PDF
Language	English
Access Restriction	Open
Content Type	Text
Resource Type	Article