Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
Attenuation compensation in PET: Incorporation of structural information from CT or MRI
Ort / Verlag
ProQuest Dissertations & Theses
Erscheinungsjahr
1994
Quelle
ProQuest Dissertations & Theses A&I
Beschreibungen/Notizen
The purpose of this dissertation is to improve image quality in positron emission tomography (PET) by applying structural information, derived from spatially correlated x-ray computed tomography (CT) or magnetic resonance (MR) images, to compensate for the effect of photon attenuation in image reconstruction. This work is motivated by the need to improve PET image quality and the availability of image registration techniques, and by the fact that CT and MR images have high image quality and provide detailed anatomic structural information. The effect of misregistration has been investigated, including the effect of geometric misregistration and the effect of reslicing. The results show that negligible errors are introduced into PET images from the procedure of image registration by the surface-matching technique. A scaling method for derivation of attenuation-coefficient maps from the resampled CT structural templates has been formulated by scaling the pixel values while considering the difference in photon energies observed in CT and in PET. Also developed were segmentation approaches in which delineation of regions of different tissue types in CT and MR structural templates was performed by different schemes, with the attenuation-coefficient maps subsequently obtained by a nonlinear transformation of pixel densities in various regions. Differences between the results obtained with those new attenuation correction methods are minimal. The new approaches have been applied to human brain and phantom studies and these results compared to the direct transmission measurement method. Quantitative evaluation of noise properties of the reconstructed images show that the new approaches reduce the %SDs of counts in image regions of interest by as much as 40-50%. The new approaches can eliminate the need for acquisition of transmission scans and are potentially applicable in routine clinical practice, with consequent improvements in clinical image quality. The work done here may be extended to applications involving other parts of the body with more sophisticated anatomic structure, as well as to applications with different imaging modalities, thus providing a basis for further research.