Evaluation von Korrelationsmodellen zwischen Surrogatsignalen und internen Featurepoints auf Basis von 4D-CTs

Lars Richter

Abstract

The aim of radiotherapy is to insert a dose into a selceted target region and to destroy tumor tissue, for example. In this process, healthy tissue should be affected as less as possible. Particularly by moving pieces of the body, this task is difficult. Especially by tumors in the lung, the respiration is responsible for large movements of the tumor. Conventionally, the dose has to be inserted into the whole movement track of the target region. The CyberKnife is a tool that can compensate the tumor movements. Because, if the position of the tumor is known at every single timepoint, the irradiation source can follow the tumor movement. Thus, the additional irradiation region will be reduced considerably. Because the tumor isn’t visible, direct tracking of the tumor is hardly possible. X-rays have to be used for estimating the tumor position. This has to be done with a high frequency to compensate the tumor movements. This would mean an additional damage to the patient, because of the required X-rays. Instead, the surface of the abdomen and of the chest is visible. Positions on these surfaces can be estimated accurately. The chest and the abdomen are moving during respiration as the tumor does. A solution of this problem gives the measurement of so called surrogate signals. This are signals that can be estimated easier and can be linked to the real position of the tumor via a mathematical modell. These mathematical modells are called correlation modells. External markers, i.e. LEDs, can be tracked precisely. As well, this applies to recognizable structures in the 4D-CT at or in the heart, at the backbone or at other points, but not necessarily for fiducials implanted in the target region. But, it isn’t always easy to find a proper surrogatesignal for a selected target region. This work concentrates on the purpose to provide a simple user enviroment for finding surrogate signals. Additionally, internal featurepoints are to be found. CT datasets that reflect the progress of a respiration cycle serve as basis of the analysis of the movements. During this work, the handling of medical data in the DICOM-format and the realtime visualisation of 4D volume data was developed. Furthermore, algorithms for the realtime localisation and tracing of distinctive points were implemented and the results were evaluated.
Original languageGerman
QualificationMaster of Science
Awarding Institution
Supervisors/Advisors
  • Bruder, Ralf, Supervisor
Publication statusPublished - 30.09.2008
Externally publishedYes

Cite this