last modified 09/23/05; WEB Comments -
webmaster@nucmed.buffalo.edu
last modified 09/23/05; WEB Comments -
webmaster@nucmed.buffalo.edu
Quick jump to: Measured Operating Specifications
The Positron Emission Tomography (PET) camera locate at the
Veterans Affairs Medical Center is a Siemens/CTI ECAT 951/31R
installed in July 1991. The scanner employs bismuth germanate
oxide (BGO) crystal detector segments arranged in an 8 crystal by
8 crystal matrix formed from a single BGO crystal block. The
individual crystal detectors have dimensions of 6.25 [mm]
transverse by 6.75 [mm] axial center to center distances and are
30 [mm] deep. Crystals are separated from each other by a 0.60
[mm] gap. Each BGO block of 64 crystals (also known as detectors)
is coupled to four photomultiplier tubes which determine event
positioning within the block. The combined block/photomultiplier
system has an approximately cubic spatial resolution of 4 [mm]
(FWHM), timing resolution of 3.4 [ns] (FWHM), coincidence windows
of 12 [ns], and energy resolution of 20% (FWHM) for 0.511 [MeV]
gamma rays. Scatter acceptance for the block through the
photopeak window is approximately 10% and 5% for external and
internal scatter events respectively. This design provides for a
stationary sampling rate (2D mode) of 55 lines of response per
[cc] throughout the sensitive volume of the tomograph.
The blocks of BGO crystals containing 64 detectors
are grouped adjacent to one another in a module consisting of
four blocks known as a bucket. Sixteen buckets positioned
adjacent to one another form a complete ring unit. This
ring unit makes up the detector groupings for acquiring 8 direct
and 7 indirect (or cross-plane) image planes for a total
of 15 image planes. With two ring assemblies contiguous to each
other in the scanner and a cross-plane formed between these two
ring, the scanner generates 31 image planes (2D mode) over an
axial field of view of 10.8 [cm] sampled by of 8,192 detectors.
Inter-plane center to center spacing is 3.375 [mm]. Transaxial
resolution in the radial direction for a 1.2 [mm] line source is
5.0, 5.7, and 7.1 [mm] (FWHM) at positions of 0.0, 10.0, and 20.0
[cm] radius respectively for a wobbled study. Transaxial
resolution under the same conditions for a stationary study are
5.8, 6.4, and 7.7 [mm] (FWHM). Axial resolution measured with a
point source is 5.0, 5.7, and 7.1 [mm] at 0.0, 10.0, and 20.0
[cm] radius respectively. The patient aperture has a diameter of
56.2 [cm]. The scanner is also configured with 15 retractable
interplane tungsten septa which, when retracted, allow 3D
acquisition of 256 planes of data. The scanner is also equipped
to run in a cardiac gated mode. Measured attenuation correction
data is obtained utilizing three rotating, retractable rod
sources filled with 68-germanium.
Coincidence events are determined by an image plane coincidence
processor and are in turn passed on to a real time sorter (RTS)
which forms sinograms in memory. After an acquisition has
completed this sinogram is transferred to disk (approx. time 1
[min]) while the RTS simultaneously accumulates the next
acquisition (if there is one) in a different area of memory. The
total memory available to the RTS is 32 megabytes (MB) with an
individual stationary acquisition (31 planes) requiring 3 MB.
Acquisitions are controlled by a 68020 microprocessor operating
on a VME bus. Image reconstruction is performed on a 66 MFLOPS
array processor and all resultant data are stored on a 676 MB
SCSI acquisition disk. This drive's data may be backed up to a
2.3 gigabyte (GB) 8mm magnetic tape drive.
An operators console consisting of a Sun SPARC 20/Unix station is
linked with the acquisition processing/storage system via an
Ethernet interface. The operators console is equipped with a
floating point processor, 32 MB memory, 2D/3D graphics
accelerator, 10" color monitor, 1.0 GB hard disk, 150 MB QIC
magnetic tape, CD-ROM, floppy drive, and a 620 MB erasable
optical drive for image archiving. Text output is obtained from a
PostScript compatible laser printer. High resolution 8x10 [inch]
film or 35mm slide capabilities are present for hard copy image
documentation.
In addition to the scanner's basic computer configuration,
additional Sun SPARC/UNIX stations for image display and analysis
are attached to the system via thin wire (10base2) Ethernet
connections.
Measured operating specifications at a glance:
Values are presented in the format "M ± S (P%)" where M is the mean of the quantity, S the associated standard deviation, and P is the percent of the mean represented by S. Values in square brackets [] below measured data indicate acceptance criteria specified by Siemens.
1) Transaxial resolution [mm] using 1.2[mm] ID needle filled with 18F.
Stationary mode [192 projections x 256 views]:
0[cm]: 1[cm]: 10[cm]: [6.40] [7.00] N.A. 5.70 ± 0.15 (2.7%) 6.37 ± 0.14 (2.1%) N.A. 5.88 ± 0.07 (1.3%) 6.07 ± 0.11 (1.8%)
Wobbled mode [384 projections x 512 views]:
0[cm] 1[cm] 10[cm] [5.50] [6.30] 5.03 ± 0.21 (4.2%) N.A. 5.58 ± 0.12 (2.1%) 5.04 ± 0.12 (2.4%) N.A. 5.19 ± 0.10 (1.8%)
2) Axial resolution [mm] using 21 gauge needle filled with 18F
For factory tests, offsets were in the horizontal direction, for site tests, offsets were in the vertical direction.
0[cm] 10[cm] [5.5] [6.6] 4.84 ± 0.33 (6.9%) 5.44 ± 0.54 (9.9%)
3) Efficiency test with 20[cm] flood phantom (LLD 250 [keV], ULD 850 [keV])
115,316.74 [counts/sec]/[µCi/ml] (total of all 31 planes) 3.00[mCi] 18F)
4) Image uniformity all planes, Stationary mode, flood phantom
2.53% ± 0.56, Avg. dev.: {4.12% ± 0.69} [15%]
5) Twelve hour repeatability test. At the factory, this test was run with the 68Ge flood phantom. Since this was unavailable for the on site tests, an alternate source was improvised utilizing the transmission rod sources and acceptance criteria were based on total net trues per plane instead of image ROI values.
All planes repeatable net trues within ±1.0% as compared to first frame's counts (without taking source decay into consideration, an effect on the order of 0.15%).
Factory tests using a 20[cm] diameter cylindrical flood phantom filled with 68-Ge gave all planes were within ±0.4% [±1.0%] for true plane counts using a 15[cm] image ROI, over the entire time interval.
6) Count Rate test with 18F flood phantom
Initial activity in the phantom was 7.39[µCi/ml]. At this concentration, the average ROI ratio (compared to final frame of test) for all planes is 95.6%±3.9% with a maximum value of 104.6% and a minimum value of 90.9%. At 5.39[µCi/ml], these values are 98.8%±2.4% with a minimum of 95.2% and a maximum of 104.7%. All plane values remain within the specified ±10% in the 5.00 to 0.10[µCi/ml] range. Dead time ranged from a factor of 5.78 to 1.03.
7) Retractable rod source background counts
With the three retractable rod sources retracted into their shielded housings, the trues rate observed (all planes) was 0.22 ± 0.13 (57%) [trues/sec]. The highest trues rate observed in any plane was 0.71 [trues/sec]. This is well within the acceptance criterion of 3.33 [trues/sec].
8) Retractable rod source repeatability test
Extending the rods, retracting them, and extending them again demonstrated repeatability within 0.03% ± 0.21% for planes 1-16, and -0.02% ± 0.13% for planes 17-31. Poisson statistics were on the order of 0.10%. All planes fell within the acceptance criterion of ±1%.
9) Retractable septa repositioning test.
10) Laser positioning
A spatial resolution test was run using a 68Ge rod source positioned at the horizontal and vertical intersections of the lasers. This test indicated that the lasers correctly defined the center of the FOV for all 31 planes to within ±3[mm]. While the two vertical positioning lasers are off slightly, with a 3[mm] between them in the center of the scanner, this misalignment is too small to cause any positioning difficulties. For the purpose of the test, the rod was placed at a location midway between the two vertical laser indicators.
last modified 09/23/05; WEB Comments -
webmaster@nucmed.buffalo.edu