3D Parametric Intensity Models for the Localization of 3D Anatomical Point Landmarks and 3D Segmentation of Human Vessel

Kein Bild zugeordnet
Autor/en:
S. Wörz
Umfang:
321
EAN/ISBN:
978-3-89838-299-1
Erscheinungsdatum:
Samstag, 30. September 2006
Band:
299
Ausgabe:
softcover
Buchreihe:
Dissertationen zur Künstlichen Intelligenz
Kategorien:
Buch
Informatik
Künstliche Intelligenz
Dissertationen zur Künstlichen Intelligenz
Englisch
Gesamtverzeichnis AKA Verlag
Preis:
55,00 €
inkl. 7% MwSt.
This thesis addresses two important problems in the field of 3D medical image analysis, namely the localization of 3D anatomical point landmarks and the segmentation of 3D tubular structures. 3D anatomical point landmarks are useful image features in a variety of applications, for example, for the registration of 3D brain images. We introduce a new approach for the localization of 3D anatomical point landmarks, which is based on 3D parametric intensity models. By fitting the intensity models directly to the image intensities we obtain a subvoxel estimate of the landmark position. Our approach has been successfully applied to accurately localize anatomical landmarks in 3D MR and 3D CT images. It turns out that our new approach significantly improves the localization accuracy in comparison to previously proposed 3D differential approaches. Concerning 3D tubular structures, the accurate segmentation of vessels from 3D MRA and 3D CTA images is crucial for diagnosis, treatment, and surgical planning. In this thesis we introduce a new approach for the 3D segmentation of vessels. This approach is based on a new 3D cylindrical intensity model, which is directly fitted to the image intensities within a 3D ROI. To segment a complete vessel, an incremental (segment-wise) estimation process based on a Kalman filter is utilized. Segmentation results are the vessel centerline and shape as well as the contrast and image blur. Our model has been successfully applied to segment vessels from 3D MRA and CTA images. The experiments show that the new model yields superior results in estimating the vessel width compared to previous approaches.