Refine
Year of publication
Document Type
- Doctoral Thesis (505) (remove)
Language
- English (262)
- German (241)
- Multiple languages (1)
- Spanish (1)
Has Fulltext
- yes (505)
Is part of the Bibliography
- no (505)
Keywords
- Pestizid (8)
- Pflanzenschutzmittel (8)
- Führung (6)
- Inklusion (6)
- Grundwasserfauna (5)
- Landwirtschaft (5)
- Modellierung (4)
- Persönlichkeit (4)
- Selbstwirksamkeit (4)
- Software Engineering (4)
Institute
- Fachbereich 7 (93)
- Fachbereich 8 (47)
- Institut für Informatik (39)
- Institut für Integrierte Naturwissenschaften, Abt. Chemie (32)
- Institut für Integrierte Naturwissenschaften, Abt. Biologie (31)
- Institut für Umweltwissenschaften (23)
- Fachbereich 5 (20)
- Institut für Computervisualistik (18)
- Institut für Wirtschafts- und Verwaltungsinformatik (14)
- Institut für Integrierte Naturwissenschaften, Abt. Physik (13)
Positron Emission Tomography (PET) is becoming more and more important in clinical routine
applications. One of the major limitations is the sensitivity to patient motion especially in the thorax
to periodic respiratory movement. Another open point of discussion is the method how to define the
tumor volume, especially when the precise knowledge of the tumor borders is important as in
radiation treatment planning. Therefore, in this work these two topics to improve quantification in
PET imaging have been addressed. First a new motion correction algorithm was implemented using
image deblurring including movement information of a 4D Computed Tomography (CT). This method,
which has the advantage of not increasing the PET acquisition time as other motion correction
techniques, was applied to phantom and patient data and showed promising result in improvement of lesion quantification. In phantom studies an improvement of up to 49% in lesion volume and in
patient studies of up to 33.3% could be demonstrated.
In the second part of this work, a new segmentation method based on textural parameters was
implemented and validated as well in phantom and patient data. In the latter a validation with
histopathological data was performed showing a very good performance of the new algorithms,
especially in larger lesions. Best result could be shown in phantom data and patient data for the
segmentation algorithm based in the local entropy.
In summary, two algorithms were implemented and validated which can improve quantification of
PET imaging furthermore