Medical image analysis using model-based optimization
Proceedings of the First Conference on Visualization in Biomedical Computing - VBC '90, Atlanta, GA, USA, 22 - 25 Mayıs 1990, ss.370-377, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Basıldığı Şehir: Atlanta, GA, USA
- Sayfa Sayıları: ss.370-377
- Boğaziçi Üniversitesi Adresli: Evet
Özet
The use of mathematical models and optimization theory is suggested as a unifying approach to solving certain automated segmentation and analysis problems in medical imaging. Within this framework, the authors present three examples that represent the use of models described by mathematical functionals and optimization techniques to solve problems that arise at three different levels of image analysis. Being able to view segmented objects and/or quantitative descriptions of an object's shape, size, and movement properties is felt to be an improved form of visualizing the anatomy for purposes of diagnosis. The first two examples are related to extracting features and ultimately segmenting outlines of objects which are useful for more clearly representing part of the anatomy, as well as for permitting further quantitative analysis of such structure. These methods are useful for many two-dimensional anatomical object recognition problems. The third example is an analysis method that is specific to the problem of tracking motion of the wall of the left ventricle of the heart. The authors present initial work describing the integration of the functionals formed from the first two levels into a hierarchy, an approach which has several computational and conceptual advantages, leading to more efficient and accurate analysis.