Biomedical Engineering Reference
In-Depth Information
occur. Such a misalignment can be explained by the fact that the CT and PET
acquisitions are not pursued under the same protocol; a cine CT acquisition
takes usually about 30 sec corresponding to about 6 respiratory cycles, while
a PET acquisition of a single bed position lasts between 4 and 12 min. As
a result, it is more likely that the patients' breathing may uctuate, both
in phase and amplitude, during the CT and PET acquisitions. Accordingly,
because of this difference between the CT images and PET emission data, the
elastic transformation parameters calculated from the three different sets of
images are not identical and when integrated in the reconstruction process;
motion-compensated images are also different (see Figure 9.6). Lamare et
al. [31] have demonstrated over eight clinical cases that, in the vast majority
of cases, it is preferable to calculate the parameters of transformation from
the PET images not corrected for attenuation rather than from attenuation-
corrected PET images or CT images. Not correcting for attenuation avoids any
potential mismatch that may exist between the CT images and PET emission
data.
The image resolution in cardiac PET is significantly degraded due to the
combination of physiological cardiac and respiratory motion [55]. With recent
combined PET/CT systems, dual-gated acquisitions simultaneously synchro-
nized with cardiac beating and respiration was proven to be feasible in both
preclinical and clinical PET/CT cardiac imaging [67, 37, 27]. Moreover the
motion of the heart due to respiration is not uniform across the left ventri-
cle [38]. To account for this observation, the use of an elastic deformation
algorithm has been proposed to realign individual dual-gated frames [32].
This study has demonstrated that, when trying to correct for respiratory
movement only by considering either the whole cardiac cycle or the diastolic
phase only, an ane or elastic motion model-based transformation applied
during the reconstruction process leads to significant improvements in overall
image qualitative and quantitative accuracy. However, when all the cardiac
and respiratory gated frames are considered independently, the motion correc-
tion technique employed has to be able to compensate for both cardiac and
respiratory motions simultaneously. In this case, the ane model shows its
limitations and the elastic model based approach obtains higher performance
in terms of contrast recovery and repositioning.
9.4 Combination of parameter identification and motion
estimation
As explained in detail in these two sections there is a plethora of applica-
tions for parameter identification as well as for motion correction approaches
in emission tomography. Both approaches were introduced as independent re-
 
Search WWH ::




Custom Search