MMS: Modular model based segmentation framework
Dr. Karl Fritscher, M.Sc.
____________________________________________________________________

Despite of the large number of innovations in terms of new technologies and strategies for the acquisition and processing of medical images, segmentation of medical images is still a challenging problem. In order to per-form efficient image segmentation in a clinical background, methods for fully automatic image segmentation have to be provided.

For this purpose a wide range of algorithms from different fields of image processing have to be adapted and new methods have to be developed. Beside of segmentation and image preprocessing algorithms, methods to generate shape models in order to initialize and/or guide the segmentation process are needed. Moreover, registration algorithms to pre-register individual images to an atlas or labeled data sets for the task of modeling are needed. For this task we are using an approach based on non-rigid registration in combination with level set segmentation and statistical deformation models.

Apart from the development of fully automated segmentation algorithms, the implementation of tools for user-friendly and intuitive semi-automatic segmentation would be able to satisfy the immediate need of clinicians for an easy to use aid in image segmentation for different clinical tasks. Therefore it is meaningful to provide also tools for semiautomatic segmentation, giving the user the possibility to manually define seed points or borders for the segmentation process.

The open source libraries “Insight Segmentation and Registration Toolkit” (ITK) and the “Visualization Toolkit” (VTK) and “Fast Light Toolkit” (FLTK) have been chosen to act as a basis for the implementation of image processing filters, visualization tasks and a general user interface (GUI), in order to develop the described algorithms in a modular C++ software framework.


Partners:  

State: ongoing (ends in Feb. 2006)

Funded by: FFF/HITT

Associated thesis:

J. Holzmeister, 4D-Segmentierung von MRI-Daten menschlicher Herzen, IBIA

G. Hribernik, 4D-Segmentierung von Endocard und Epikard menschlicher Herzzyklen, IBIA

M. Osl, Erstellung eines Tools zur Transformation von 3D-Labelbildern unter Verwendung von Surface Meshes, IBIA

H. Plieger, Implementierung von Werkzeugen zur Visualisierung medizinischer Bilddaten in ein C++-Bildverarbeitungs­framework, IBIA

V. Spiteller, Implementierung von binären Operatoren in ein C++-Bildverarbeitungs­framework, IBIA

Publications:

K. D. Fritscher, R. Pilgram, R. Leuwer, C. Habermann, A. Müller, R. Schubert, Analyzing inter-individual shape variations of the middle ear cavity by developing a common shape model based on medial representation, CARS 2004, Chicago

K. D. Fritscher, R. Schubert, A software framework for pre-processing and level-set segmentation of medical image data, SPIE Medical Imaging
13.02.2005-18.02.2005, San Diego, CA;

K. D. Fritscher, R. Pilgram, R. Schubert, Automatic cardiac 4D segmentation using level sets, Funtional Imaging and Modeling of the Heart (FIMH), 2. 6. – 4. 6. 2005, Barcelona, Spain

K. D. Fritscher, R. Pilgram, R. Schubert, Automatic 4D endocardium segmentation using hierarchical registration and model guided level set segmentation, Computer Aided Radiologie and Surgery (CARS), 22. 6. – 24. 6. 2005, Berlin

K. D. Fritscher, Development of a software framework for pre-processing and level-set segmentation of medical image data, Master-Thesis, IBIA

 

Startseite

Aktuelles

Forschung

Publikationen

Kooperationen

Lehre

Mitarbeiter


zurück zur UMIT

zurück zu den Instituten