OVERVIEW OF THE ACR MRI ACCREDITATION PHANTOM

Geoffrey D. Clarke, Ph.D.

University of Texas Southwestern Medical Center at Dallas
email: [email protected]

˝
INTRODUCTION
˝

˝
The ACR magnetic resonance accreditation phantom (ACR MRAP) has been
designed to examine a broad range of instrument parameters. These include:

• Geometric Distortion
• Spatial Resolution
• Slice thickness and position
• Interslice Gap
• Estimate of Image Bandwidth
• Low Contrast Detectability
• Image Uniformity
• Signal-to-Noise Ratio (SNR)
• Physical and Electronic Slice Offset
• Landmark
This document contains phantom specifications and materials and a brief
description of image analysis. For a more complete description of the ACR MRI
Accreditation Phantom, you may order the publication, Phantom Test Guidance
for the ACR MRI Accreditation Program, from the American College of
Radiology.

PHANTOM SPECIFICATIONS

The ACR MRI accreditation phantom is constructed of acrylate plastic, glass,

and silicone rubber. Ferromagnetic materials have been excluded. The unit is a
cylinder 20.4cm in diameter by 16.5cm in length. Internal dimensions are 19.0
cm diameter by 15.0 cm in length. The compact design allows placement in
axial coronal, or sagittal orientation in almost all MRI head coils. Thus, tests can
be conducted in all three major planes. There is a reference line down one side
of the phantom.

The phantom is filled with 10 millimolar (mmol) nickel chloride solution containing
sodium chloride (45 mmol) to simulate biological conductivity. The contrast vial
contains 20 mmol nickel chloride and 15 mmol sodium chloride solution providing

GEOFFREY D. CLARKE 1 MRI PHANTOMS & QA TESTING

a difference in T1 and T2 values. Actual values will depend on the field strength
in use and the temperature of the phantom.

The resolution insert on one end of the phantom consists of three matrices of
holes in an 11mm thick bar. Hole diameters are 1.1mm, 1.00mm, and 0.9mm.
The spaces between the holes are equal to the respective hole diameters. This
insert is used to test limiting in-plane spatial resolution.

Two counter-descending wedges are found at this end of the phantom. They
each contain a 1 cm slit. The wedges form two ramps of test solution which
descend at a 1:10 ratio to permit accurate measurement of slice thickness.

The grid insert toward the center of the phantom is a 10 by 10 array of squares
144 mm on a side and 10 mm thick. It is used for placing the diagonal lines in
the geometric distortion tests. The nominal interior diameter of the phantom is
190 mm

GEOFFREY D. CLARKE 2 MRI PHANTOMS & QA TESTING

 45 wedges Low
Contrast
Disks

100
mm Array of Squares 15
cm

Resolution Insert
45 wedges

19 cm

Figure 1. Saggital localizer view of ACR MRI

Accreditaion Phantom with several prominent
landmarks labeled.

Four low-density contrast disks are located on the far end of the phantom. They
consist of thin sheets of polycarbonate plastic 0.002, 0.004, 0.006, and 0.008
inch in thickness. Holes of different diameters have been cut into the disks.
Partial volume contributions of both the fill solution and these membranes
produce slight variations in signal strength which may be used to visually assess
the scanner’s ability to distinguish low contrast objects.

Two sets of paired 45o wedges are located on the top and bottom of the
phantom. Each pair is 2 cm in length with the center of intersections at 1 cm
from either end. The distance between the intersection points of the paired
wedges is 100mm (see Figure 1). The wedges are used to precisely measure
physical and electronic slice offsets. The paired wedges can also be used to
evaluate small interslice gaps.

GEOFFREY D. CLARKE 3 MRI PHANTOMS & QA TESTING

IMAGING PROTOCOLS
˝

˝
• 1. Saggital Localizer: TR = 200 ms, TE = 20 ms, 256x256 matrix, 25 cm
FOV, 10 mm slice thickness, NSA=1, single saggital slice (52 seconds).
Landmark on the center reference line.

• 2. T1 Weighted Multislice Study: TR = 500 ms, TE = 20 ms, 256x256

matrix, 25 cm FOV, 12 slices, 5 mm slice thickness with 5 mm gap in
between, NSA=1(2.2 minutes). Begin at center of front set of wedges
as seen in localizer.

• 3. T2 Weighted Multislice Multiecho Study: TR = 2000 ms, TE1 = 20

ms, TE2 = 80 ms, 256x256 matrix, 25 cm FOV, 12 slices, 5 mm slice
thickness with 5 mm gap in between, NSA=1(8.5 minutes). Begin at
center of front set of wedges as seen in localizer.

Deviations from imaging protocols:

Some MRI systems, particularly older ones, will not be capable of performing
the scans as indicated above. In such cases the ACR allows the applicants
to submit images acquired using their standard imaging protocols.

GEOFFREY D. CLARKE 4 MRI PHANTOMS & QA TESTING

 12
11
10
9
8
7
6
5
4
3
2
1

Figure 2. Saggital localizer image with slice positions

for axial T1-weighted and T2-weighted images. The
white arrow along right-hand side indicates distance
between 45 degree wedge crossings for length of
phantom measurement.

Image Analysis
Bandwidth/Chemical Shift

The chemical shift insert is composed of square structures, one containing

10mmol nickel chloride solution and the other vegetable fat . These square
structures are arranged catty corner but will appear shifted toward or away from
each other, depending on the direction of the chemical shift. Bandwidth (BW)
can be assessed by measuring the chemical shift in millimeters, dividing the FOV
by this and multiplying the result by 3.5 ppm of the magnet’s operating
frequency.

Landmark
˝

˝
Landmark accuracy can be checked by examining the bars in the first slice. If
they are of equal length and if the acquisition was started from the center of the
first set of paired wedges (see saggital localizer) landmark is correct.

GEOFFREY D. CLARKE 5 MRI PHANTOMS & QA TESTING

GEOFFREY D. CLARKE 6 MRI PHANTOMS & QA TESTING
Slice Thickness

The center insert in slice #1 should be used to measure slice thickness. The two
slits in the large wedges ascend and descend at a 1 to 10 rate. A slice taken
through these wedges will show lines of signal where the slit (filled with solution)
is encountered. Measuring the length of these bright lines in millimeters,
averaging (to remove errors from slight misalignment of the phantom), and
dividing by 10 will give the slice thickness. The edges of the signal region fade to
black, and the distance measurement should be made from a point of half-
maximum signal on each end of each wedge.

Resolution

The resolution portion of the phantom contains three matrices of holes with
spacings of 1.1mm, 1 .0 mm, and 0.9mm. The spaces between the holes are
the same distance as the respective hole diameter. The holes are offset slightly
to account for positioning differences. Slice #1 (5mm,25cm FOV, 256x256
matrix) should show individual 1.1mm and 1.0 mm holes but only a blur for the
0.9 mm sets. A magnified image, carefully adjusted for window and level to
resolve the smallest part of the resolution pattern, should evaluated. This test
should be run without any data shaping filters (e.g., Fermi, Hanning, Hamming,
etc.).

Figure 3.

Image of
Slice #5 with array of
squares. Grey arrows
indicate the positions
of the measurements
for determinig percent
geometric distortion.

GEOFFREY D. CLARKE 7 MRI PHANTOMS & QA TESTING

Geometric Distortion:
Visual inspection is often sufficient to detect severe warping or stretching of the
grid of squares (Image #3). The distance calculation function on the console
should be used to verify that the inner diameter of the phantom is accurately
measured. The inner diameter of the phantom is 190mm. Total distance across,
from top to bottom, and along each diagonal should be 190 mm. An angle along
the side of the grid should read 90 degrees. Measurement directly on film can
be used to detect geometric distortion in the matrix or laser camera. Horizontal,
vertical, and diagonal measurements across the phantom diameter should give
the same result

N
Figure 4.

Image of Slice #7 used

for signal-to-noise
measurements.
Signal is obtained from
mean value of large
circular ROI. Maximum
and minimum signal
values should also be
N SIGNAL
obtained from within this
N
ROI. Noise is obtained
from the standard
deviation of the signal in
the background ROI’s,
depicted by the ellipses
marked “N”. Noise ROI’s
should be placed above,
below, to the right and to
the left of the phantom
N image.

Signal-to-Noise
The first few slices of acquisition #2 contain a small circle from the vial of 20
mmol nickel chloride solution. A region of interest (ROI) of approximately 2.5
cm2 is placed on the circle and the average pixel value is recorded (Figure 3).
Outside the phantom image (where no signal should be generated) the average
pixel value in the corners is recorded in four ROI’s of 15-20 cm2 (background).
The standard deviation of this value is also taken as it reflects the noise inherent
in the system. The solution value minus the average of the background value is
a measure of signal. This is divided by the average of the standard deviation of
the background to yield SNR. A similar analysis should be conducted on the T2-
weighted echo in acquisition #3.
Radio Frequency (RF) Uniformity

GEOFFREY D. CLARKE 8 MRI PHANTOMS & QA TESTING

RF uniformity can be assessed by analyzing slice #7 (Figure 4), taken from the
flood section of the phantom. Over the center ROI of the image (approximately
210 cm2) determine the SNR using the same technique as described for slice
#1. With the windows width turned down to minimum value determine the
brightest and darkest 1 cm2 areas within the center ROI by varying the level.
Record the SNR for these 1 cm2 areas. Subtracting each from the average SNR
and dividing by the average SNR gives the percent peak-to-peak variation in
signal across the flood image.

Figure. 5

Four slices with

low-contrast
detectibility
inserts. More
Slice #8 Slice #9 of the sets of
three holes are
visualized as one
Wedge
goes from slice #8
length to slice #11. Note
difference that in slice #11
the wedge length
difference is used
to indicate slice
position offsets.

Slice #10 Slice #11

Interslice Gap
Interslice gap can be checked by comparing the "bars" from the small paired
wedges on each end of the phantom. The wedges are each 2cm long, set at an
angle of 45o, and are separated by 100mm. By starting the multi-slice at the
center of the first set of wedges, the bars will appear equal in that slice. If this is
not true there may be an error in the internal landmarking system of the MRI
scanner. Eleven slices at 5 mm skip 5 mm should move 100 mm which is the
distance to the center of the second wedges. This image of slice #11 (Fig. 5),
should also have equal bars. The difference divided by two is the error in the
total of these slices expressed in mm. Thus, if slice thickness is correct and the
overall distance shift is correct, interslice gap must also be accurate.
Low Contrast Detectability

GEOFFREY D. CLARKE 9 MRI PHANTOMS & QA TESTING

The thin polycarbonate membranes in the low contrast inserts displace a small
amount of fill solution, reducing signal in proportion to their thickness. When
imaged with a relatively thick slice (5 mm) partial voluming yields differences in
signals between areas where the membrane exists and where the are holes in it.
Four variations in thickness produce a contrast range from 2% to 8%.

MRI PHANTOM AND QA TEST REFERENCES

Rice SO. Mathematical analysis of random noise. (Parts 1 & 2) Bell Sys Tech J 1944; 24:46-
156.
Rice SO. Mathematical analysis of random noise.(Parts 3 & 4) Bell Sys Tech J 1945; 25:282-
332.
Hoult DI, Richards RE. The signal-to-noise in the nuclear magnetic resonance experiment. J
Magn Reson 1976; 24:71-85.
Hoult DI, Lauterbur PC. Signal-to-noise in nuclear magnetic resonance zeugmatography. J
Magn Reson 34:425-433.
Edelstein WA, Bottomley PA, Pfeifer LM. A signal-to-noise calibration procedure for NMR
imaging systems. Med Phys 1984; 11:180-185.
Henkelman RM. Measurement of signal intensities in the presence of noise in MR images.
Med Phys 1985; 12:232-233.
Maudsley AA. Sensitivity in Fourier imaging. J Magn Reson 1986; 68:363-366.
Hendrick RE. Sampling time effects on signal-to-noise and contrast-to-noise ratios in spin-echo
MRI. Magn Reson Imag 1987; 5:31-37.
Kraft KA, Fatouros PP, Clarke GD, Kishore PRS. An MRI phantom material for quantitative
relaxometry. Magn Reson Med 1987; 5:555-562.
Selikson M, Fearon T. Averaging error in NMR slice profile measurements. Magn Reson Med
1988; 7:280-284.
Kaufman L, Kramer DM, Crooks LE, Ortendahl DA. Measuring signal-to-noise ratios in MR
imaging. Radiology 1989; 173:265-267.
Parker DL, Gullberg GT. Signal-to-noise efficiency in magnetic resonance imaging. Med Phys
1990; 17:250-257.
Price RR, Axel L, Morgan T, Newman R, Perman W, Schneiders N, Selikson M, Wood ML,
Thomas SR. Quality assurance methods and phantoms for magnetic resonance imaging.
AAPM Report #28. Med Phys 1990; 17(2):287-295.
Och JG, Clarke GD, Sobol WT, Rosen CW, Mun SK. Acceptance testing of magnetic
resonance imaging systems: Report of AAPM nuclear magnetic resonance task group #6.
Med Phys 1992; 19(1):217-227.
Bakker CJG, Moerland MA, Bhagwandien R, Beersma R. Analysis of machine-dependent and
object–induced geometric distortion in 2dft mr imaging. Magn Reson Imag 1992; 10:597-
608.
Frayne R, Holdsworth DW, Gowman LM, Rickey DW, Drangova M, Fenster A, Rutt BK.
Computer-controlled flow simulator for mr flow studies. J Magn Reson Imag 1992; 2:605.
McGibney G, Smith MR. An unbiased signal-to-noise ratio measure for magnetic resonance
images. Med Phys 1993; 20:1077-1078.
Hyde RJ, Ellis JH, Gardner EA, Zhang Y, Carson PL. MRI scanner variability studies using a
semi-automated analysis system. Magn Reson Imag 1994; 12(7):1089-1097
Gudbjartsson H, Patz S. The Rician distribution of noisy mri data. Magn Reson Med 1995;
34:910-914.

GEOFFREY D. CLARKE 10 MRI PHANTOMS & QA

TESTING