Biomedical Engineering Reference
In-Depth Information
where the signal of the analyzed voxel of the current gate starts to match the signal
of a randomly gated image. The randomly gated image is created based on random
(motion independent) trigger signals and represents thus the random noise.
Another approach for an improvement of the gated images is presented in [
24
]
by Büther et al. The idea is to attach external markers to the patient to guarantee a
sufficiently strong signal which can be used by their data-driven gating technique.
Consequently, the proposed method is also applicable to patients with low global
uptake.
4.3
Influence of Motion on PET Attenuation Correction
In oncological PET imaging, using CT or MRI to generate PET attenuation maps,
respiratory motion may introduce artifacts at the position of the diaphragm due to
μ
values derived from incorrect respiratory phases. Therefore, there is a specific
need to solve this registration problem in hybrid PET imaging. No matter whether
gating or reconstruction-based correction methods are being used, the respiratory
phase at which a corresponding CT or MRI images has been acquired needs to
be known accurately. Apart from various techniques for predicting the respiratory
phase, such as intrinsic or extrinsic markers, navigators, etc., there has been another
elegant way of generating
-maps for proper attenuation correction. With the advent
of Time-Of-Flight (TOF)-PET in clinical PET/CT scanners, an old idea [
103
]of
jointly reconstructing emission and transmission data within an integrated iterative
reconstruction algorithm became reality, called Maximum Likelihood reconstruc-
tion of Attenuation and Activity (MLAA) [
62
,
113
]. Using this approach, a
μ
-map is
reconstructed that is, by definition, registered to the corresponding emission image
obviating the need for additional attenuation estimations. In addition, truncation
problems of CT and MRI may be solved with this approach leading to more accurate
and quantitative PET images [
102
].
μ
4.4
Future Advances Using PET/MRI
One big issue in PET-based motion correction is the relatively low information
content in the PET images. Motion can only be estimated in regions of sufficiently
high tracer uptake which may limit intrinsic motion estimation approaches. A global
motion estimation and correction is thus hardly achievable. The clinically relevant
structures are usually given by these high uptake regions, which makes PET-based
motion correction practicable in many applications. However, small lesions might
be invisible in the uncorrected PET image due to motion blurring and reliable
estimation of motion is hardly possible. As such lesions are of great clinical
relevance, the idea is to make them visible again by deriving motion fields
from auxiliary measurements. As mentioned in Sect.
1.5.1
, the required motion
Search WWH ::
Custom Search