TY  - THES
AU  - Lintermann, Andreas
TI  - Surface Reconstruction of the Human Nasal Cavity from CT-Data for Fluid Mechanical Analysis of Breathing Problems
PB  - RWTH Aachen University
VL  - Diplomarbeit
CY  - Aachen
M1  - FZJ-2019-06123
SP  - 161
PY  - 2008
N1  - Diplomarbeit, RWTH Aachen University, 2008
AB  - In this diploma thesis, the accuracy of three-dimensional surface reconstructions of the human nasal cavity from Computer Tomography (CT) volume images is analyzed. For this purpose, a reconstruction pipeline is developed, allowing the extraction of the nasal cavity surface, which then can be used to carry out flow simulations to analyze breathing problems. The accuracy is examined by generating an artificial CT-image from a reference surface of the nasal cavity, approximating a real CT-image by applying filters and subsequently extracting a surface with the reconstruction pipeline. Error metrics are introduced to examine the distance between the reference and the generated surface giving information on the accuracy of the reconstructed surface. After the extraction of the raw surface from CT-data, a post-processing is necessary because of aliasing effects. The surface is smoothed with Laplacian or windowed sinc function smoothing. It is shown that the latter smoothing method allows a better approximation to the original surface without shrinking the surface as occuring in Laplacian smoothing. The curvature of the reconstructed surface is studied to obtain a measure of smoothness. The error produced under increasing CT-image resolution, using the surface obtained from the generated image, is also investigated. As expected, the error decreases under increasing the resolution. To examine the error of artifacts in CT-images, noise of a certain period and amplitude is su- perimposed to the CT-data, a surface is extracted and the error is calculated. Additionally, the volumetric change of the nasal cavity is correlated to this error, allowing an evaluation of the influence of artifacts. The results show that increasing noise is linearly correlated to a volumetric change of the nasal cavity and the surface error.
LB  - PUB:(DE-HGF)10
UR  - https://juser.fz-juelich.de/record/867484
ER  -