Download Evolution for Bone

GE Healthcare
Evolution for Bone™
Collimator-Detector Response
Compensation in Iterative SPECT
Imaging Reconstruction Algorithm
Version 1.0
Introduction
The quality and quantitative accuracy of SPECT reconstructed
images are affected by noise in the projection data, resolution
degradation caused by the collimator-detector response
(CDR) function of the imaging system, and photon attenuation
and scatter in patient’s body. Recently, significant progress
has been achieved in the development of model-based
corrective SPECT image reconstruction methods that include
correction for these image quality-degrading factors.
The corrective image reconstruction package (also referred to
as Resolution Recovery) developed by JHU group provides
improved SPECT images by including physical models of the
imaging process into iterative image reconstruction algorithm.
In particular, it provides accurate 3D models of the collimatordetector response functions for variety of clinical applications.
This results in significant enhancement in image quality and
potential improvement in diagnostic properties. Clinical
application of these techniques are available and can be
applied to reduced time SPECT acquisitions with equal or
better image characteristics when compared to standard
reconstructions (16).
Collimator-detector response in
SPECT
Collimator-detector blur is the main factor affecting the
resolution and noise properties of nuclear medicine images. In
reconstructed SPECT images, these effects are strongly
affected by the applied reconstruction algorithm and its
parameters.
In SPECT acquisition with parallel collimation, images of a
point source located at different distances from the collimator
fully describe (CDR) function for specific collimator design,
photon energy and energy window parameters, and distance
of the imaged object from collimator surface, as
demonstrated in figure 1.
Figure 1. SPECT acquisition geometry and collimator-detector response.
CDR consists of four main components: intrinsic response
(system without collimator) and the geometric, septal
penetration and septal scatter components of the collimator
parameters. The geometric component is the portion of the
CDR resulting from photons by passing through the collimator
holes without interaction with the collimator. The septal
penetration component describes portion of the CDR due to
photons reaching the detector after passing through one or
more septa without scattering. The scatter component is due
to photons that scatter in the septa and are then detected.
For the majority of clinically used collimator designs, the last
two components are generally significant for only mediumand high-energy imaging (see fig.2).
The response functions are spatially invariant in planes
parallel to the face of the collimator and that the intrinsic
response is approximately spatially invariant in the detection
plane.
Figure 3. Scheme of iterative reconstruction process.
Compensation requires that a table of CDR functions at all
relevant distances to collimator surface be pre-calculated and
retrieved, or computed on the fly, during reconstruction. Onthe-fly calculations involve using the theoretical predicted
collimator geometric response for an equivalent round hole
shape [1, 2], and ignore penetration and septal scatter effects.
These approximations work - well for conventional hexagonal
shaped holes and low-energy isotopes, such as Tl and Tc99m.
Figure 2. Examples of CDR function and its components.
For each combination of acquisition system,
radiopharmaceutical and particular acquisition protocol, the
collimator detector response function provides the probability
that a photon emitted from any point of the imaged object
will contribute to a pixel of the resulting image.
A CDR compensation technique was developed at the
University of North Carolina Chapel Hill (UNC) and Johns
Hopkins University (JHU) [3, 4, 5]. In fast rotation based
iterative reconstruction implementations, CDR compensation
is achieved by convolving the projected ray with
corresponding Point Spread Function (PSF) in the course of
projection and back-projection operations (Fig.4)
Accurate predictions of the geometric response function for
various collimator designs were derived [1, 2]. Penetration
and scatter components of the collimator detector response
function can be obtained and analyzed using Monte-Carlo
simulation methods.
Compensation for collimatordetector response in iterative
reconstruction
By including an accurate model of collimator-detector
response function in an iterative SPECT reconstruction
algorithm, the blurring effect may be included in the iterative
reconstruction process, resulting in improved special
resolution.
Figure 4. Modeling of collimator-detector response in a rotation based
projector. Adapted from: E.C. Frey and B.M.W. Tsui, Collimator-detector
response compensation in SPECT
In order to incorporate CDR into the reconstruction process,
the following information is utilized during the course of image
reconstruction:
• Collimator design parameters : holes length and diameter,
septa thickness
• Detector characteristics : intrinsic resolution, crystal
thickness, collimator-detector gap
• Acquisition parameters: center-of-rotation to collimator
face distances for every projection view acquired.
This information for a variety of different collimators is stored
in an additional look-up table which is part of the
reconstruction package.
Effect of collimator detector
response compensation on
reconstructed images
data when no compensation is applied. On the other hand,
image noise level tends to amplify with iteration number.
Taking into account the differences in the iterative
reconstruction process, with and without CDR, applicationspecific optimization of acquisition and processing
parameters is essential to successful utilization of corrective
reconstruction method [12, 13, 14, 15 ] (Fig.6)
It was demonstrated (Fig.6) that application of corrective
image reconstruction methods results in improved uniformity
of myocardium wall in cardiac SPECT and specifically CDR
compensation provides better recovery of lesion contrast in
variety of clinical applications [8, 16].
The effect of CDR compensation on reconstructed images has
been studied from qualitative [7], quantitative [8, 9, 10], and
clinical task-based [11, 13, 15] prospective.
It has been found that CDR corrected images demonstrate not
only improved resolution and signal-to-noise ratio, but also
lower noise variability in reconstructed images (Fig. 5). It has
also been shown that the noise texture with CDR
compensation changes, exhibiting wider correlations than
non-corrected images, resulting in improved resolution
without corresponding increase in noise. CDR is quite sensitive
to incorrect modeling of the CDR function or the use of
incorrect acquisition parameters. It is important that these
parameters are accurately passed from the camera to the
iterative reconstruction process.
Figure 6. Reconstructed clinical Tc-99m myocardium SPECT images. Top line attenuation correction only applied, bottom line – attenuation and CDR
compensation. Numbers mark number of updates (product of iterations
number and subsets number) applied. From: E.C. Frey and B.M.W. Tsui,
Collimator-detector response compensation in SPECT
Mathematical and human observer studies demonstrated
improvement in cold and hot lesion detectability with
application of CDR compensation [15, 16].
Clinical Application
CDR has been implemented on the Xeleris workstation for use
in the Bone SPECT and Whole Body SPECT protocols. The
application of CDR in shorter acquisition times has been
shown to demonstrate equivalent or improved image quality
over standard, full time acquisitions (16).
Image improvement of half time CDR reconstructed SPECT
over full time standard OSEM iterative reconstruction is
demonstrated in (fig. 9).
Figure 5. Hot sphere phantom images. Left: schematic presentation of insert
locations; middle: slice of conventional reconstruction, right: slice of corrective
reconstruction with collimator-detector response compensation. No post
filtering was applied.
Significant improvement of the resolution properties requires
more iterations of the reconstruction method with
compensation than are usually recommended for the same
Figure 8. Changes in Area under ROC curve (AUC) across
acquisition/processing combinations for selected 10mm 2:1 contrast lesions –
results from Channelized Hotelling Observer study.
CDR
OSEM
Figure 7. Examples of reconstructed image slices of Tc-99m bone SPECT
containing 10 mm 6:1 contrast lesions; simulated data.
Figures 7 and 8 show results from an investigation of image
properties where application of CDR compensation gave rise
to studies of potential usefulness of this compensation in
shorter than usual clinical acquisition time to ensure nondegraded diagnostic quality [16, 17].
Figure 9. Demonstrated improvement in WB SPECT half time acquisition over
standard OSEM iterative reconstruction process
Conclusion
Detailed description of the effect of the collimator detector
response on properties and diagnostic quality of SPECT
images is a subject of extensive research. So far, significant
evidence has been accumulated that diagnostic quality of NM
images is substantially improved by incorporation of
collimator-detector system modeling into iterative image
reconstruction.
References
1. Metz, C.E., The geometric transfer function component for
scintillation camera collimators with straight parallel holes.
Phys. Med. Biol., 1980. 25(6): p. 1059-1070.
2. Tsui, B.M.W. and G.T. Gullberg, The Geometric TransferFunction for Cone and Fan Beam Collimators. Physics in
Medicine and Biology, 1990. 35(1): p. 81-93.
3. Tsui, B.M.W., et al., Implementation of simultaneous
attenuation and detector response correction in SPECT. IEEE
Transactions on Nuclear Science, 1988. 35(1): p. 778-783.
4. Tsui, B.M.W., et al., The importance and implementation of
accurate three-dimensional compensation methods for
quantitative SPECT. Phys Med Biol, 1994. 39(3): p. 509-530.
5. Tsui, B.M.W., et al., Characteristics of reconstructed point
response in three-dimensional spatially variant detector
response compensation in SPECT, in Three-Dimensional
Image Reconstruction in Radiology and Nuclear Medicine, P.
Grangeat and J.-L. Amans, Editors. 1996, Kluwer Academic
Publishers. p. 509-530.
6. Wilson, D.W. and H.H. Barrett, The effects of incorrect
modeling on noise and resolution properties of ML-EM
images. Nuclear Science, IEEE Transactions on, 2002. 49(3):
p. 768-773.
7. Tsui BMW, Zhao XD*, Frey EC and McCartney WH.
Quantitative SPECT: Basics and Clinical Considerations.
Seminar in Nuclear Medicine, Vol. XXIV, No 1 (January), pp
38-65, 1994
8. Pretorius, P.H., et al., Reducing the influence of the partial
volume effect on SPECT activity quantitation with 3D
modelling of spatial resolution in iterative reconstruction.
Physics in Medicine and Biology, 1998. 43(2): p. 407-420.
9. Kohli, V., et al., Compensation for distance-dependent
resolution in cardiac-perfusion SPECT: impact on uniformity
of wall counts and wall thickness. Nuclear Science, IEEE
Transactions on, 1998. 45(3): p. 1104-1110.
10. Pretorius, P.H., et al., Comparison of detection accuracy of
perfusion defects in SPECT for different reconstruction
strategies using polar-map quantitation. Ieee Transactions
on Nuclear Science, 2003. 50(5): p. 1569-1574.
11. Narayanan, M.V., et al., Human-observer receiveroperating-characteristic evaluation of attenuation, scatter,
and resolution compensation strategies for Tc-99m
myocardial perfusion imaging. Journal of Nuclear Medicine,
2003. 44(11): p. 1725-1734.
12. Sankaran, S., et al., Optimum compensation method and
filter cutoff frequency in myocardial SPECT: A human
observer study. Journal of Nuclear Medicine, 2002. 43(3): p.
432-438.
13. Frey, E.C., K.L. Gilland, and B.M.W. Tsui, Application of taskbased measures of image quality to optimization and
evaluation of three-dimensional reconstruction-based
compensation methods in myocardial perfusion SPECT. IEEE
Transactions on Medical Imaging, 2002. 21(9): p. 1040-1050.
14. Gifford, H.C., et al., LROC analysis of detector-response
compensation in SPECT. IEEE Transactions on Medical
Imaging, 2000. 19(5): p. 463-473.
15. He, X, et al., A mathematical observer study for the
evaluation and optimization of compensation methods for
Myocardial SPECT using a phantom population that
realistically models patient variability. IEEE Transactions on
Nuclear Science, 2004. 51(1): 218-224
16. Volokh, L., et al., Efficacy of corrective reconstruction with
collimator detector response compensation for short Tc99m bone SPECT acquisition in a bone lesion detection
task./ Abstract presented at SNM Meeting, 2005/
17. Keidar, Z. et al., Half-time bone SPECT acquisition Assessment of a new Collimator Detector Response (CDR)
reconstruction algorithm / Abstract submitted to RSNA
Meeting, 2005 /
18. Frey, E.C., Tsui, B.M.W., Collimator-Detector Response
Compensation in SPECT. in: Quantitative Analysis in Nuclear
Medicine Imaging. Springer 2005
GE Healthcare
Waukesha, WI
U.S.A.
www.gehealthcare.com
© 2005 General Electric Company – All rights reserved.
GE Healthcare, a division of General Electric Company.
General Electric Company reserves the right to make
changes in specifications and features shown herein,
or discontinue the product described at any time without
notice or obligation. Contact your GE Representative for
the most current information.