A Probabilistic Framework for Point-Based Shape Modeling in Medical Image Analysis

1. Front Matter

   Pages i-xxiii

   PDF

2. Introduction

   • Heike Hufnagel

   Pages 1-5

3. Current Methods in Statistical Shape Analysis

   • Heike Hufnagel

   Pages 7-25

4. A Generative Gaussian Mixture Statistical Shape Model

   • Heike Hufnagel

   Pages 27-55

5. Evaluation of the GGM-SSM

   • Heike Hufnagel

   Pages 57-80

6. Using the GGM-SSM as a Prior for Segmentation

   • Heike Hufnagel

   Pages 81-113

7. Conclusion

   • Heike Hufnagel

   Pages 115-120

8. Back Matter

   Pages 121-147

   PDF

In medical image analysis, major areas such as radiotherapy, surgery planning, and quantitative diagnostics benefit from shape modeling to facilitate solutions to analysis, segmentation and reconstruction problems.
Heike Hufnagel proposes a mathematically sound statistical shape model using correspondence probabilities instead of 1-to-1 correspondences. The explicit probabilistic model is employed as shape prior in an implicit level set segmentation. Due to the particular attributes of the new model, the challenging integration of explicit and implicit representations can be done in an elegant mathematical formulation, thus combining the advantages of both explicit model and implicit segmentation. Evaluations are performed to depict the characteristics and strengths of the new model and segmentation method.

The dissertation has received the Fokusfinder award 2011 by the Innovationsstiftung Schleswig-Holstein (ISH), the Basler AG and Philips Medical Systems.

• Medical Image
• Model-based Segmentation
• Probabilistic Correspondences
• Segmentierung
• Shape Modeling

Heike Hufnagel works as a research scientist at the Institute of Medical Informatics, University of Luebeck, Germany. She holds a PhD in natural sciences.

Policies and ethics