Biomedical Engineering Reference
In-Depth Information
methods have first been developed for use in SPECT systems [40] [45] [30]
and have been translated to PET. The main difference between these meth-
ods is the way the energy windows are chosen. In the dual energy windows
approach for PET [37], the standard photopeak window is set between 380 keV
and 850 keV, an additional, lower energy window is set between 200 keV and
380 keV. Both windows will contain scattered and unscattered events (unscat-
tered events that do not deposit their full energy in the detectors will appear
in the lower window). The data measured during the scan in both windows is
then used in combination with predetermined line source scan data with and
without scatter medium to determine the scatter distribution in the standard
photopeak window. The true, unscattered events in this window can then be
found by subtracting the smoothed scatter distribution from the total event
rate. The dual energy window method showed good results in terms of accu-
racy and ease of implementation in a study comprising numerical simulations,
hardware phantom scans, and clinical patient data [90].
The triple energy method makes use of two additional energy windows
besides the usual photopeak energy window. These windows are overlapping
and have the same upper level setting of 450 keV. The ratio of events in the
two windows for the scanned object and for a homogeneous cylinder defines
a calibration that can be used to determine the scatter rate in the photopeak
window [81]. Other energy windowing methods comprise the usage of multiple
energy windows (typically 16 16 = 256 for coincidence events; multispectral
method [7]); however, this mode of acquisition may be limited by the scanner
and may thus not be installed easily on PET systems.
Recently, energy-based scatter estimation in PET has been realized by
taking detailed energy information available in list mode data into account
during image reconstruction [77].
5.5.2 Analytical methods
Analytical methods aim to determine the scatter distribution from the
emission data acquired in the photopeak window. This is done using a simple
analytical model for the scatter distribution. Two of these approaches based
on deconvolution and tail fitting shall be described here in greater detail.
As in the energy windowing methods described above, the acquired raw
data g 0 in emission tomography can be described as a sum of both unscattered
events g u and scattered events g s :
g 0 = g u + g s : (5.21)
In deconvolution-based scatter correction methods, g s is now modelled
as a spatial convolution of the unscattered distribution g u with a spatially-
dependent scatter function f and a scatter fraction k that both describe the
scatter properties of the scan:
g s = k
ยท
(g u f) :
(5.22)
 
Search WWH ::




Custom Search