Skip to Main content Skip to Navigation
Theses

Segmentation par contours actifs en imagerie médicale dynamique : application en cardiologie nucléaire

Abstract : In emission imaging, nuclear medicine provides functional information about the organ of interest. In transmission imaging, it provides anatomical information whose goal may be the correction of physical phenomena that corrupt emission images. With both emission and transmission images, it is useful to know how to extract, either automatically or semiautomatically, the organs of interest and the body outline in the case of a large field of view. This is the aim of segmentation. We developed two active contour segmentation methods. They were implemented using level sets. The key point is the evolution velocity definition. First, we were interested in static transmission imaging of the thorax. The evolution velocity was heuristically defined and depended only on the acquired projections. The segmented transmission map was computed w/o reconstruction and could be advantageously used for attenuation correction. Then, we studied the segmentation of cardiac gated sequences. The developed space-time segmentation method results from the minimization of a variational criterion which takes into account the whole sequence. The computed segmentation could be used for calculating physiological parameters. As an illustration, we computed the ejection fraction. Finally, we exploited some level set properties to develop a non-rigid, non-parametric, and geometric registration method. We applied it for kinetic compensation of cardiac gated sequences. The registered images were then added together providing an image with noise characteristics similar to a cardiac static image but w/o motion-induced blurring.
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00506987
Contributor : Eric Debreuve <>
Submitted on : Thursday, July 29, 2010 - 1:03:47 PM
Last modification on : Wednesday, October 14, 2020 - 4:23:27 AM
Long-term archiving on: : Thursday, November 4, 2010 - 10:03:37 AM

Identifiers

  • HAL Id : tel-00506987, version 1

Collections

Citation

Eric Debreuve. Segmentation par contours actifs en imagerie médicale dynamique : application en cardiologie nucléaire. Traitement du signal et de l'image [eess.SP]. Université Nice Sophia Antipolis, 2000. Français. ⟨tel-00506987⟩

Share

Metrics

Record views

838

Files downloads

520