Level set techniques for structural inversion in medical imaging
Résumé
Most biological bodies are structured in the sense that they contain quite well-defined interfaces between regions of different types of tissue or anatomical material. Extracting structural information from medical or biological images has been an important research topic for a long time. Recently, much attention has been devoted to quite novel techniques for the direct recovery of structural information from physically measured data. These techniques differ from more traditional image processing and image segmentation techniques by the fact that they try to recover structured images not from already given pixel or voxel-based reconstructions (obtained, e.g., using traditional medical inversion techniques), but directly from the given raw data. This has the advantage that the final result is guaranteed to satisfy the imposed criteria of data fitness as well as those of the given structural prior information. The 'level-set-technique' plays an important role in many of these novel structural inversion approaches, due to its capability of modeling topological changes during the typically iterative inversion process. In this text we will provide a brief introduction into some techniques that have been developed recently for solving structural inverse problems using a level set technique.