Biomedical Engineering Reference
In-Depth Information
In the case of respiratory movements (PET), it is common to use external devices
that generate a
trigger
signal for protocols based on gated acquisition [
62
]. Some
of these devices are the pressure sensors, spirometers, temperature sensors and also
dedicated video cameras.
3.3
Strategies for Motion Correction
The techniques that aimed at reducing the effects of movement were formerly
categorized by Baimel et al. in three groups [
43
]:
gating
techniques, computational
techniques and analog circuit techniques. The analog circuit techniques include
those that apply electronic circuits to the outputs of the position of the gamma
camera and correct the detected events based on a centroid calculation [
43
,
44
].
With the development of computers, these techniques were naturally replaced by
others that were more efficient.
Currently, the gating and the computational techniques continue to be consid-
ered, though new categories have been proposed to better classify the correction
algorithms that have been devised.
Gating methods should not be considered as a motion compensation technique
in the strict sense, since in fact, it is assumed the presence of a particular class of
motion which is inevitable but trackable. Generally, the motion is periodic such as
the heartbeat. Hence, instead of trying to devise a way of eliminating the motion
(with post processing computation), the acquisition is performed at precise instants
allowing to obtain images at different phases of movement. The drawback that has
to be overcome is the small number of counts (due to dosimetric considerations)
that are usually obtained for each image. To solve this situation the motion cycle is
divided in a certain number of intervals and data is accumulated on each of these
intervals cycle after cycle (Fig.
13
). It is, therefore, indispensable the existence of a
way to trigger each new cycle - in the case of cardiac imaging it is used the R wave
of electrocardiogram.
One of the most common forms of performing motion compensation in clinical
routine (for dynamic acquisitions) is to detect motion by visualization of the images
and then to remove the images with motion. Despite being quick and easy to perform
implies a loss of number of counts per pixel which increases noise. An efficient
alternative to the image elimination is the manual alignment of images which will
ensure an improvement of the final image [
63
]. However, this technique is time
consuming, cumbersome and user dependent and is therefore worse than others.
In the case of planar scintigraphy (e.g. renogram [
64
]) the motion correction
is performed after the detection of movement, by aligning the images with proper
interpolation methods. This simple approach allows only the correction of either
translational movements that occur in the projection plane or rotation about a
perpendicular axis to the plane of projection.
Tomography requires more complex methods that involve the techniques of
image reconstruction. The refinement of the methods is clear in the literature: firstly
