Motivation

Intensity modulated radiotherapy tries to realize a high dose level in a specified target volume in order to destroy cancer tissue. It is absolutely necessary that the precise place of the target volume is known during the treatment time which typically will last over weeks. An application of doses suitable for destruction of cancer cells to healthy tissue due to incomplete knowledge of the patients geometry might lead to serious complications. Therefore, there is a strong need for a periodic update of the patients geometry, and, in consequence of the physical treatment setup, during the treatment time or in between the treatment fractions.
Changes in the patients geometry are in general complicated to realize and to model: simple shifts or rotations are not sufficient. More generally, a rigid matching is not appropriate. There is need for an elastic matching.

Methods

There are basically two classes of adaptive matchings: methods of the first class work on features extracted from the geometry data. By searching for corresponding areas that have strong similarities with regard to the features the motion of the tissue can be determined indirectly. The precision of these methods depend heavily from the precision with which the features can be realized. Methods of the second class try to generate the unknown transformation directly from the geometry data. Depending on the structure of picture typically only parts of the translation field can be constructed. In order to obtain complete information about the transformation a sufficiently rich local structure of the pictures is required.

Example

The example represents a changing organ geometry during a treatment time. Comparison of original and current organ arrangement shows a significant change of form, orientation and size of the target volume.

Stand: 01.08.03

Contact:

	Dr. habil. Karl-Heinz Küfer Tel.: +49 (0)631/36681-13