Keywords
collimator - gamma camera - planar imaging - SIMIND - tungsten - spatial resolution
- Wolfmet
Introduction
A collimator in scintillation camera is mandatory for creating an image of the in
vivo radiopharmaceutical distribution. It usually consists of a large number of parallel
holes in a block of high-attenuating material, and the purpose of these holes is allow
only those photons traveling parallel to the collimator holes to pass through and
interact in the crystal and contribute to the image. The dimensions of the holes (hole
diameter and length) and the septum between them together with the collimator material,
for a specific photon energy, determine the spatial resolution and the system sensitivity
and thus have an impact on the image quality.[1]
[2]
[3]
[4]
[5]
[6]
[7] Lead (Pb), added with a small fraction of antimony (Sb), is the most commonly used
collimator material for full-sized clinical scintillation cameras, and the collimators
are, based on this material, optimized to imaging photons ranging from 140 to 364 keV.
However, several studies have suggested alternative materials, such as tungsten (W),
gold (Au), uranium (U), and platinum (Pt), all of which have better attenuating properties
for photons with reduced collimator scatter and septal penetration.[2]
[8]
[9]
[10]
[11]
[12] For example, the attenuation coefficients, μ, for tungsten at 140.5 and 364 keV
are 36.23 and 4.34/cm, respectively. This can be compared with 26.86 and 3.13/cm for
Pb.[13] However, the increased cost and the increased weight of such collimators have made
these more useful for smaller collimators, such as the pinhole collimator where the
actual aperture can be made by the materials mentioned above but the surrounding collimator
housing is made by Pb.[14]
[15] For examples, the new clinical CZT-based multidetector single-photon emission computed
tomography (SPECT) systems, including GE StarGuide (GE Healthcare, Inc) and Veriton
(Spectrum Dynamics, Inc), are equipped with tungsten parallel-hole collimators. These
systems are, however, based on smaller detectors than a full-sized camera and the
collimators are optimized for 140 keV. Also, the dedicated cardiac cameras, D-SPECT
(Spectrum Dynamics, Inc) and Discovery NM 530 (GE Healthcare, Inc), are both based
on tungsten collimators.
A drawback with tungsten is its high melting point and that it cannot be cast like
Pb-based collimators and the metal is also difficult to machine. This makes pure tungsten
collimators more expensive than Pb collimators. An alternative is using tungsten in
alloys (with nickel, iron, and/or copper), which can then be precisely milled or drilled
with a diamond drill or machined using electric discharge machining (EDM). However,
these techniques are also very expensive, and also complex shapes like strongly tilted
pinholes, loftholes, or pinholes with small opening angles cannot be easily produced.[2] However, pure tungsten can be of interest as a collimator material for magnetic
resonance (MR) compatible SPECT systems because tungsten alloys can often contain
magnetic materials.[16]
More recently, there have been new developments in collimator production techniques,
which make the fabrication of more complex collimator designs possible.[17] These include direct three-dimensional (3D) printing of metals and “cold casting”
of W-based composite materials. This opens new possibilities for more tailored collimator
designs that otherwise would be impossible or very expensive to construct by conventional
methods such as casting hot lead (micro-cast) or folding lead foils (micro-linear).[17]
[18]
[19]
[20]
[21]
[22] W-based collimators with a mass density of 18.56 g/cm3 and a mean deviation of 35 µm were recently reported for a tungsten parallel-hole
collimator with a hole size of 0.525 mm, septal thickness of 0.150 mm, and hole length
of 25 mm.[15]
Könik et al have modeled two multipinhole (MPH) collimator designs with 9- and 16-pinhole
apertures in a 2-cm-thick tungsten alloy (tungsten 90% and Cu 10%; mass density: 18.2 g/cm3) to study primary, scatter, and penetration characterizations of parallel-hole and
pinhole collimators for I-123 SPECT and found that the MPH collimators provide superior
count performance in terms of high primary counts, low penetration, and low scatter
counts compared with the Pb-based parallel-hole and single pinhole collimators.[23]
For full-size cameras, there are today three families of collimators, namely, low-energy
(LE) collimators, medium-energy (ME) collimators, and high-energy (HE) collimators.
LE collimators can be provided as optimized for low energy high sensitivity (LE-HS),
for spatial resolution (LEHR), or as low energy all purpose collimator (LEAP), and
all collimators are mainly used for imaging photons with energies of 140.5 keV (Tc-99m)
or below. The ME collimators are generally optimized for In-111 (245 keV) and Ga-67
(300 keV). HE collimators are mainly optimized for I-131. A LEHR collimator can be
useful for I-123 imaging because the principal photon energy, 159 keV, is close to
140.5 keV, but the decay also includes emission of photon with energies above 400 keV,
which can contribute negatively to the image quality. This is also the case for I-131
that emits two HEs (637 and 364 keV) that both make a significant contribution to
the counts in the main energy, 364.5 keV.[24]
The aim of this study was therefore to study a potential benefit of using W-based
metals as a possible collimator material to reduce the problems mentioned earlier.
The study was conducted using Monte Carlo simulations to compare images, obtained
from W-based material, with conventional Pb-based collimators regarding system spatial
resolution, sensitivity, and image contrast. The study was based on planar imaging
since the main purpose was to investigate the influence of the collimator material
so the result should be translated to SPECT, since reconstructions are based on planar
projection images. Since the aim was not to find an optimal collimator design, regarding
collimator hole properties (diameters, lengths, and septum thicknesses), the study
was based on commercial collimators but with different collimator materials.
Materials and Methods
The SIMIND Monte Carlo program was used to simulate a Siemens scintillation camera,
equipped with a 0.375-inch-thick NaI(Tl) crystal and conventional HE, ME, and LEHR
parallel-hole collimators ([Table 1]).[25]
[26] The energy resolution was defined to 9.5% at 140 keV and the intrinsic spatial resolution
was set at 3.5 mm. A single slab of Pyrex with a thickness of 6 cm was used as a substitute
for photomultiplier tubes (PMTs), electronics, and other compartments to include potential
contribution due to backscattering of those emitted photons with higher energy at
the principal photopeak energy.[27] The energy-dependent cross-sectional values for the collimator materials, shown
in [Table 1], were generated from the XCOM database.[13]
Table 1
Details on the materials used as a parallel-hole collimator
|
Material
|
Composition
|
Mass density (g/cm3)
|
Effective atomic number
|
|
Pb
|
Pb (98%)
Sb (2%)
|
11.25
–
|
81.63
–
|
|
W
|
W (100%)
|
19.25
|
74.0
|
|
Wolfmet
|
W (97.0%)
Ni (1.5%)
Fe (1.5%)
|
18.5
|
72.6
|
The choice of the composition of the W alloy was based on Wolfmet HE-397 metal,[28] and this specific alloy, with properties described in [Table 1], was selected from several types of alloys that have been described in a previous
study.[29] The specifications of the collimators used in this work were taken from the vendor's
data sheet.
The weights of the W-based collimators presented in [Table 2] were estimated by scaling the Pb weight with the fraction of the W and Wolfmet mass
density to the Pb mass density, respectively. A small overestimation is expected here
since not all parts of a collimator are made by lead.
Table 2
Specifications of the hexagonal parallel-hole collimators
|
Collimator type
|
|
Specifications
|
HE
|
ME
|
LEHR
|
|
No. of holes
|
8,000
|
14,000
|
148,000
|
|
Hole diameter (mm)
|
4.00
|
2.94
|
1.11
|
|
Hole length (mm)
|
59.70
|
40.64
|
24.05
|
|
Septum thickness (mm)
|
2.00
|
1.14
|
0.16
|
|
Weight[a] of Pb (kg)
|
124.7
|
63.5
|
22.1
|
|
Weight of W (kg)
|
211.7
|
107.8
|
37.5
|
|
Weight of Wolfmet (kg)
|
200.1
|
101.9
|
35.5
|
a Weight of Pb collimators taken from the vendor's data sheet.
The isotopes I-123 and I-131 were selected for this study since both have decay schemes
that include emission of HE photons that are known to cause penetration problems and
contribute to degradation of image quality. Lu-177 was selected due to its common
use. It has some HE photons in the decay, but these have a relative low abundance.
Tc-99m was selected as a common isotope for imaging that do not have emissions in
the decay of photons with HE than the principal energy.
Simulations were made for four types of the sources, namely (1) a point source from
the lower surface of the collimator; (2) a uniform circular source distribution with
an 11-cm radius that included six cold circular areas with radii equal to the spheres
in the NEMA IC phantom (30.0, 18.5, 14.5, 11, 8.5, and 5 mm); and (3) a hot source
geometry, similar to the cold source, but the sources with a relative activity concentrations
of 2, 3, 4, 6, 8, and 12; and (4) a two-dimensional (2D) model of the hot rods of
the Plexiglass Deluxe ECT (model ECT/DLX/P) bar phantom in a cold background. All
sources were located at a distance of 10 cm from the lower surface of the camera.
A main 15% energy window width was set over the photopeak energies of 140.5, 159,
208.4, and 364.5 keV for Tc-99m, I-123, Lu-177, and I-131 isotopes, respectively.
A complete decay spectrum was used for each isotope in the simulations. The matrix
size was set to 256 × 256 with 1.12-mm pixel size (zoom factor 4) for the point source
simulations and 128 × 128 with 2.24-mm pixel size (zoom factor 2) for the other source
distributions.
The system's spatial resolution and the sensitivity in the whole field of view (cps/MBq)
were determined from the simulated planar images of the point source. A Gaussian fitting
function was used to calculate the full width at half maximum (FWHM) of the point
spread function (PSF) from two types of images representing (1) the total counts in
the images and (2) the counts created only from geometrically collimated photons to
determine the impact on septal penetration. The image contrast was calculated for
the two largest circles in the uniform activity phantom (cases with cold sources).
To calculate the contrast, the average counts, C̅roi
, were calculated from a circular region of interest (ROI) with radii equal to a fraction
of 0.6 of the physical radii of the ROIs to reduce the partial volume effect. A background
ROI was positioned in the center of the source with a radius that will not interfere
with the other ROIs.
Contrast = [1 = C̅roi
/C̅bkg
] × 100%
Results
PSFs from the events for the isotopes in the simulated images, created by geometrically
collimated photons and from all photons, are shown in [Fig. 1] for the HE, ME, and LEHR collimators of Pb, W, and Wolfmet. The results of the calculation
of the FWHM, the system sensitivity, and the calculated fraction of geometrically
collimated counts to the total counts are tabulated in [Table 3]. The vendor's specification for the spatial resolution of HE, ME, and LEHR collimators
were 13.4 mm (I-131), 12.5 mm (Ga-67), and 7.5 mm (Tc-99m), respectively. When compared
with the simulated data of spatial resolution, the FWHM from the simulated Pb collimators
showed a good agreement. The vendor's specifications for septal penetration were 1.5,
1.2, and 3.5%, which were generally lower than what we found in our study. This difference
can be subjected to how the septal penetration is defined. The definition of septal
penetration event in SIMIND is such as when the position of the photon at the exit
surface of the collimator is outside the physical boundaries of the current entrance
hole plus one septal thickness.
Fig. 1 Point spread functions (PSFs) obtained for the three collimators and for I-131, Lu-177,
I-123, and Tc-99m isotopes from a point source located 10 cm from the lower surface
of the collimator. The dashed line represents counts from geometrically collimated photons along the profile and the
solid line represents all registered counts. ME, medium energy.
Table 3
Results of the spatial resolution for total counts (T), the geometrical collimated
counts (G), the system sensitivity in units of cps(T)/MBq and the G/T fraction of
the counts for a 10-cm source-to-collimator distance
|
Radionuclide
|
Collimator
|
FWHM(T)
|
FWHM(G)
|
Sensitivity
|
1-G/T
|
|
I-131
|
HE (Pb)
|
6.2
|
6.2
|
40.1
|
28%
|
|
ME (W)
|
13.1
|
12.2
|
69.7
|
45%
|
|
ME (Wolfmet)
|
13.1
|
12.2
|
77.4
|
50%
|
|
Lu-177
|
ME (Pb)
|
11.7
|
11.4
|
11.7
|
6%
|
|
ME (W)
|
11.5
|
11.3
|
11.3
|
5%
|
|
ME (Wolfmet)
|
11.6
|
11.3
|
11.4
|
5%
|
|
Lu-177
|
LEHR (Pb)
|
7.4
|
7.2
|
42.7
|
84%
|
|
LEHR (W)
|
7.3
|
7.1
|
17.4
|
61%
|
|
LEHR (Wolfmet)
|
7.3
|
7.1
|
20.9
|
67%
|
|
I-123
|
LEHR (Pb)
|
7.0
|
6.9
|
111.3
|
34%
|
|
LEHR (W)
|
6.9
|
6.8
|
91.4
|
21%
|
|
LEHR (Wolfmet)
|
6.9
|
6.8
|
94.0
|
23%
|
|
Tc-99m
|
LEHR (Pb)
|
6.9
|
6.8
|
89.9
|
7%
|
|
LEHR (W)
|
6.8
|
6.7
|
86.1
|
4%
|
|
LEHR (Wolfmet)
|
6.8
|
6.8
|
86.8
|
4%
|
Abbreviations: FWHM, full width at half maximum; HE, high energy; ME, medium energy;
ROI, region of interest.
When comparing HE Pb with ME W/Wolfmet collimators for I-131 imaging, it can be seen
that there is a higher spatial resolution in W-based collimators. The difference in
FWHM between W and Wolfmet is very little, but with an increase in penetration for
the Wolfmet, as expected due to the lower mass density. However, when viewing the
bar images, shown in [Fig. 2], it is hard to see any visible difference. This suggests that an ME W-based collimator
could be used to obtain the same image quality as the HE Pb collimator. The sensitivity
for the W-based ME collimator is higher and this is not due to septal penetration
only since the peaks in the PSF also are higher. Thus, an advantage can be found in
both spatial resolution and sensitivity when using ME W-based collimators but at the
expense of an increased septal penetration, as can be seen from the tails of the PSF
in [Fig. 1].
Fig. 2 Planar images of the bar sources at 10 cm from the collimator for the three collimators
and the I-131, Lu-177, I-123, and Tc-99m isotopes. HE, high energy; ME, medium energy.
For Lu-177 imaging, the results showed that there was little advantage of using W-based
ME collimators. The PSFs are almost identical in shape and the FWHM values differ
very marginally. The penetration problem is thus a small problem, which indicated
that ME Pb collimator works fine with Lu-177 and may actually be suboptimal with the
208-keV energy. When imaging Lu-177 with an LEHR W-based collimator and compared to
an ME Pb collimator, there was a major improvement in spatial resolution. For LEHR
Pb collimators, however, a large fraction of penetration as can be seen, as shown
in [Table 3] and [Fig. 1]. These collimators are therefore not recommended for imaging with 208 keV.[30] However, a potential W-based LEHR collimator significantly reduced the penetration
and improved the spatial resolution. The sensitivity is also higher due to the extended
number of holes. There are, however, penetration effects that might need to be accounted
for, but the improvement in image quality is clearly seen from the bar phantom simulation,
shown in [Fig. 2]. The difference in sensitivity between W and Wolfmet was marginal.
For I-123 imaging, an LEHR W-based collimator provide a higher spatial resolution
and the negative effect of septal penetration can be seen as being small. The sensitivity
is slightly lower than that for an LEHR Pb collimator, but the fractions of events
from geometrically collimated photon are higher. From [Table 4], we can see that the image contrast for I-123 with W-based collimators is increased.
For Tc-99m, there was little advantage with LEHR W collimators in the present configuration.
The penetration was small for Pb-based collimators and there was no improvement in
spatial resolution and image contrast when using W-based collimators.
Table 4
Image contrast for the two circles in [Fig. 4] with a radius of 30 and 14.5 mm (the largest and the third largest)
|
|
30-mm ROI
|
14.5-mm ROI
|
|
Isotope
|
Collimator
|
Pb
|
W
|
Wolfmet
|
Pb
|
W
|
Wolfmet
|
|
I-131[a]
|
HE/ME
|
83%
|
73%
|
59%
|
76%
|
64%
|
59%
|
|
Lu-177
|
ME
|
100%
|
100%
|
98%
|
97%
|
97%
|
98%
|
|
LEHR
|
51%
|
70%
|
52%
|
36%
|
58%
|
52%
|
|
I-123
|
LEHR
|
77%
|
84%
|
81%
|
73%
|
82%
|
81%
|
|
Tc-99m
|
LEHR
|
99%
|
100%
|
99%
|
98%
|
100%
|
99%
|
Abbreviations: HE, high energy; ME, medium energy; ROI, region of interest.
a ME collimators is used for W-based materials.
[Fig. 3] shows the cold-cased phantom with six cold circles. The calculated contrast between
the ROIs inside the two largest circles is shown in [Table 4]. For I-131, the contrast is lower for W-based collimators than that for Pb, indicating
the contribution of the surrounding activity to the counts in the ROI due to the septal
penetration. Visually the differences are subtle, but the smallest circle with the
W-based collimators appears to be smaller due to a better resolution. The contrast
levels with the hot-cased phantom with six hot circles were seen the same as with
cold circles ([Fig. 4]). As shown in [Figs. 3] and [4], Lu-177 provided better contrast for both cold and hot circles with LEHR W-based
collimators than that for Pb, as with I-123. However, for Lu-177, there were little
contrast differences with the ME collimators. The contrast for the hot-cased phantom
with I-131 was comparable with ME W-based and HE Pb collimators.
Fig. 3 Planar images of uniform distribution with the six cold spheres at 10 cm from the
collimator for the three collimators and the I-131, Lu-177, I-123, and Tc-99m isotopes.
HE, high energy; ME, medium energy.
Fig. 4 Planar images of the six hot spheres with relative activity concentrations of 2,
3, 4, 6, 8, and 12 for the three collimators and the I-131, Lu-177, I-123, and Tc-99m
isotopes. HE, high energy; ME, medium energy.
Discussion
This study investigated the potential positive effects of the system on functional
parameters including spatial resolution, sensitivity, and image quality when using
tungsten and tungsten alloys as collimator material of a gamma camera system. The
study was conducted for planar imaging, but since SPECT is based on planar projection
images, acquired around the patient, it is assumed that any improvement seen for planar
images could translate to improvement in SPECT. Of course, there are steps in the
reconstruction process that affect, for example, the spatial resolution such as noise
regularization methods and number of iterations. On the other hand, SPECT with iterative
reconstruction has a good possibility to compensate for the degradation of spatial
resolution which is related to the events from collimator septal penetration if information
of such events is included in the collimator response function (CRF) correction, that
today are common in clinical software systems. It is, in principle, also possible
to correct for penetration effects in planar images by creating filter functions that
account for the addition of penetration to the PSF. Sjögreen et al have shown improvement
in image quality in I-131 whole-body imaging by using a Wiener-like restauration method
using Monte Carlo–based scatter kernels obtained from simulation of point sources
at various depth in a water phantom.[31] This method has also been applied for 90Y bremsstrahlung imaging.[32]
For all cameras, there is a maximum weight that the gantry can carry. In this study,
we simulated a camera system that can carry an HE Pb collimator of 126 kg. We did
not include HE W-based collimators as, according to [Table 2], we assumed that these would be too heavy. The weights of the ME and LEHR W-based
collimators were, based on our rescaling method, such that these could potentially
be carried by the system. This implies that there are also possibilities to optimize
the ME and LEHR collimators for better imaging properties than the current configurations
based on Pb.
Current ME Pb collimator configurations are probably not optimal for Lu-177, which
is an isotope that is increasingly used for therapy of neuroendocrine tumors with
Lutetium Lu 177 dotatate that binds to somatostatin receptors on tumors and metastasized
prostate cancer using a prostate-specific membrane antigen (PSMA) labeled to Lu-177.
A reduced septal thickness is possible because of the better attenuation properties
of W-based collimators, which could be reduced allowing for more holes with a related
increase in sensitivity or spatial resolution. The use of a W-based collimator, optimized
for 208 keV and spatial resolution, could improve image-based dosimetry procedures
and lead to more accurate absorbed dose estimates compared with the present Pb-based
ME collimators.
Our previous study on the image quality of Tc-99m and when using LE collimators of
W and Wolfmet alloys showed the superiority of W alloy collimators compared with the
Pb and Pb-Sb.[29] So the results are consistent with the findings in the present study for ME and
HE collimators with I-131 and Lu-177 imaging. Finally, more studies with other hole
geometries and energy ranges with the materials are needed for the optimal spatial
resolution and septal penetration values.
Conclusion
This study has shown that the use of potential collimators made of tungsten and Wolfmet
tungsten alloy could provide an improved spatial resolution and image contrast by
reducing septal penetration compared with conventional Pb-based collimators. The results
indicate that the development of new collimators with better resolution as compared
with the current HE and ME collimators could improve imaging with I-131 and Lu-177.
A W-based LEHR collimator could improve the image quality for I-123 imaging and still
be useful for Tc-99m imaging.
