INHALT & PFAD:
Für den Fall, dass Ihr Computer das entsprechende Format nicht angezeigen kann, können Sie sich hier das Dokument als unformatierte Textausgabe ansehen.
(Bitte haben Sie dafür Verständnis, dass gewisse Dokumente nicht als Textalternative zur Verfügung stehen.)
Einladung IUB-Vortrag Dr. Preusser 20080213
School of Engineering and Science
Wednesday, February 13, 2008
5:30 pm, Reimar LÃ¼st Hall, Conrad Naber Lecture Hall
Tea at 5:00 pm - All are welcome!
Image Based Computing for the Planning of
Dr. Tobias Preusser
Center of Complex Systems and Visualization (CeVis)
University of Bremen
Radio-frequency (RF) ablation is a minimally invasive therapy for primary tumors and metastases in the human
body. It is applied to treat lesions in the kidneys, the liver, the lungs, and also in bones. A probe containing electrodes
is placed into the malignant tissue. Upon connecting the probe to a generator, an electric current flows through the
tissue. Consequently, the Ohm-resistance of the tissue causes the development of heat, which destroys the malignant
The success of the treatment heavily depends on the local structure of the vascular system, and a variety of patientspecific properties of the tissue. Unfortunately, tissue properties of individual patients are not exactly known. Values
used in simulations are mostly taken from ex-vivo experiments with animal organs. In the interest of the success of
the therapy a thorough planning must be made, which yields an optimal position and orientation of the probe, and
which takes these important patient-specific properties into account.
The talk focusses on various aspects of a numerical support for the planning of an RF-ablation. A system of partial
differential equations (PDEs) is described, which models the underlying bio-physical processes.
Due to the complicated boundaries and internal structures of the organs under consideration, it is particularly
challenging to solve those PDEs with moderate numerical effort. We use a technique, which bridges the gap between
image processing and simulation with PDEs: Using level-set functions to define the geometry and utilizing concepts
from image processing and scientific visualization, the approach inherits the efficiency of image processing
To take into account the uncertainty and errors associated with material parameters, we consider those to have
distributions over certain ranges of possible values. Thus, the deterministic PDE model is generalized to a stochastic
PDE model. An optimal probe placement can be obtained from the minimization of objective functions, which
involve expectations of the stochastically distributed temperature profile.
Invitation IUB-Talk Dr. Preusser 20080213 Einladung IUB-Vortrag Dr. Preusser 20080213