Transcript Prostate - University of Michigan

P21-20
High Resolution PET Instruments for Prostate Imaging
Neal Clinthorne1, Stan Majewski2, Karol Brzezinski3, Sam S. Huh4, Jessie Carr4, Zhewei Chen4, Emma Salomonsson4, Aashay Yande4
1Dept.
Radiology University of Michigan, Ann Arbor, MI 2Center for Advanced Imaging, Dept. Radiology, West Virginia University, Morgantown, WV
3IFIC/CSIC University of Valencia, Spain, 4Dept. Biomedical Engineering, University of Michigan, Ann Arbor, MI
Background
Prototype Development
Probe Sensitivity and Anticipated Performance
As an imaging modality in prostate cancer, positron
emission tomography (PET) allows the possibility of
“engineering” radiotracers to follow specific metabolic
pathways or indicate the presence of biomarkers
associated with disease. However, our considerable
investment in tracer development cannot be used to full
advantage if PET instrumentation is not capable of
imaging small prostatic lesions well. Presently, PET
can detect prostatic lesions of ~8mm diameter. This
due to strong attenuation of the annihilation radiation C-11-choline PET
prostate cancer.
from the prostate, intrinsic resolution limits of the
scanner, and patient motion during scanning.
20000
18000
Events Detected
16000
BGO Only Probe 1
10000
BGO+LSO Probe 2
8000
BGO Only Probe 2
4000
2000
0
0
5
10
15
20
25
30
35
Axial Extent (# 5mm rings)
of
4 mm
18
6 mm
14
12
1.0 mm Probe
1.5 mm Probe
2.0 mm Probe
10
Noise Advantage over 4mm Scanner
image
Monte Carlo simulations using an anthropormorphic
phantom, probe, and 80cm dia. PET ring showed that ~12–
60% more coincidence events from the prostate were
detected depending on detector size (upper right). Plot
below right shows effect on performance over PET ring
alone in terms of noise advantage vs. resolution for only
12% additional efficiency and for probes having 1mm,
1.5mm, and 2mm resolution. A noise advantage of 10 is
equivalent to 102 or 100x the number of events.
20
8
6
4
8 mm
14
10
8
6
4
20mm
45mm
40mm
85mm
0
1
2
3
4
5
6
7
8
9
10
Reconstructed resolution (mm FWHM)
Mechanical Testing
Originally, aluminum was the material of choice for
constructing prototypes; however, it was determined that
ABS plastic was considerably less expensive and had the
required strength. (Yield strength 42 MPa, highest stress is
~22 MPa.) Seams will be sealed with parylene, which is
biocompatible and a good electrical and moisture barrier.
New silicon photomultipliers and arrays of LYSO scintillators
are enabling technologies for constructing miniature, highperformance PET detectors. Shown at right is a diagram of a
probe detector having depth-of-interaction resolution
developed by our collaborators at West Virginia University.
For details on the detectors, please visit poster P21-17
12
Although the technologies associated with the basic
concept and detector are absolutely essential, there are
many additional ingredients necessary for testing with
human subjects. These include the mechanical design of
the probe housing, considerations for temperature control,
mounting of electronics, and tracking the position and
orientation of the probe, which may move during the PET
acquisition. To address these issues, we enlisted a team
from a two-semester design course in the Department of
Biomedical Engineering at Michigan. Shown immediately to
the right is the current design, which houses the detector,
motion sensors, and EM shielding (see author for 3D
model).
2
Detector Technology
16
Relative 
BGO+LSO Probe 1
12000
6000
Resolution v. Noise Tradeoff
One way to improve resolution is to use an image
reconstruction method that models—and then
undoes—blurring inherent in the measurements.
While this can work to some extent, it inevitably
increases the level of noise in reconstructed
images. There is an intrinsic “resolution-noise”
tradeoff associated with each imaging system.
Shown at right are resolution-noise tradeoff
curves for three PET scanners having different
intrinsic resolution (4mm, 6mm, and 8mm
FWHM). Notice how quickly the noise level rises
as one attempts to work at resolution better than
supported by the intrinsic resolution.
14000
30mm
2
0
1
2
3
4
5
6
7
8
9
10
Demonstration of High Resolution PET Concept
Resolution (mm FWHM)
Motion Tracking Tests
Better Resolution Through Better Measurements
Conventional PET ring
Excellent
resolution
close to probe
The basic high resolution PET concept has been demonstrated with a partial BGO PET ring (100cm
dia.) and two high resolution silicon detectors (1.4mm res.) located close to a small (4cm) field-ofview. Reconstructed resolution phantom images shown are from BGO-BGO, BGO-Si, and Si-Si
events (left-to-right). Spot diameters are 4.8, 4.0, 3.2, 2.4, 1.6, & 1.2mm. Note the “lever-arm”
resolution improvement over BGO-BGO reconstruction with BGO-Si data (as for the prostate probe).
0.5
Residual (mm)
Modest
resolution
Residual plot
0.4
0.3
0.2
z axis
0.1
y axis
0
Demonstration of the “Limited-Angle” Imaging Probe
-0.1
0
2000
4000
6000
8000
10000
12000
Number of steps
PET ring
Low-noise and accurate position-tracking is essential for maintaining high spatial
resolution. Shown above right is a setup used for measuring linearity and resolution of the
Ascension 3D Trakstar pulsed magnetic system as well as position, linearity, and residual
plots. Range measured is considerably larger than will be needed in practice. Resolution
(rms) is ~0.3mm, which will not significantly degrade performance of probe.
High resolution endorectal
PET detector (“probe”)
6
External probe
insert
Object position
5
Resolution (mm FWHM)
Another method for improving spatial resolution is to use
a high-resolution detector probe as an add-on to a
standard PET scanner. While “ring-ring” coincidences
will still support modest resolution, “probe-ring”
coincidences will have excellent resolution near the high
resolution probe detector. Moreover, effects induced by
positron acolinearity, photon depth-of-interaction, and
patient motion are significantly reduced for activity within
the prostate. Photon attenuation is also reduced. Shown
at right is a plot of the intrinsic spatial resolution as a
function of distance from the probe for a conventional
PET ring having 6mm intrinsic resolution and probes
having 1mm, 2mm, and 3mm intrinsic resolution. Note
that the improvement can be significant even at >5cm
from the probe face.
4
3
3 mm
2
2 mm
1
0
Probe resolution = 1mm FWHM
0
5
10
15
20
25
30
Distance from probe face (cm)
35
40
The panel above showed use of full-angle tomographic data while the
prostate imaging probe will only acquire limited-angle high-resolution
data. To evaluate effects, an external probe acquiring limited-angle
data was simulated and results are shown above. External ring had
4mm and probe had 1mm resolution. Images left-to-right are
preliminary reconstructions from ring-ring, probe-ring, and probe-ring
and ring-ring coincidences combined. Combined performance should
exceed that of probe-ring alone as work progresses. Images at left
show effects of adding limited angle BGO-Si data to poorer resolution
BGO-BGO data alone (from above testbed). Although artifacts are
present, resolution improvement is noticeable.
Conclusions
Work to develop a high resolution prostate imaging add-on is proceeding rapidly in parallel
along three fronts: detector design, probe theory and image reconstruction, and housing
design for testing with human subjects. Progress toward our goal of better PET imaging of
the prostate has been significant with development of several suitable high-resolution
detectors, demonstration of the high resolution PET and PET imaging probe concepts, and
design of a prototype housing suitable both for initial tests with phantoms and for eventual
testing with prostate cancer patients. Work remains on interfacing the probe to a clinical
PET/CT instrument; however, PET manufacturers have shown interest in the technology,
which should be useful for guiding biopsy (especially when combined wth MRI or
ultrasound images).
This work was partially supported by the U.S. Army Medical Research and Materiel Command CDMRP Prostate Synergistic Idea grant W81XWH-09-1-0420.