High resolution for improved visualization on

new mobile X-ray systems Optima XR240amx

X-ray Engineering white paper

Authors Abstract

Matthieu Guillard, MSc, Image Quality Senior Engineer GE’s new mobile X-ray system Optima XR240amx is
Ping Xue, PhD, Image Quality Principal Engineer introducing the FlashPad HD wireless cassette-size
German Vera, MSc, Image Quality Engineering Manager digital flat panel detectors with an 100µm pixel pitch
and an improved pixel design. It provides a higher
spatial resolution and a higher detection efficiency at
all frequencies and all dose levels when compared to
Optima XR220amx, which justify the use of smaller
pixel size.

Characterization of these improvements is realized

using industry standard detector performances metrics
(DQE, MTF, limiting resolution) and a study of Contrast
Detail phantom for which the detectability of small
objects is especially increased.

Helix, the digital image processing of Optima

XR240amx system, is optimized to take advantage of
the additional high resolution content in the raw image
for the display of a sharper processed image compared
with Optima XR220amx. The benefit is illustrated with
clinical images for extremities exam and line placement
assessment for which high resolution is a key element
of clinical diagnosis.

Note: Optima XR200amx and Optima XR220amx

systems can be upgraded to Optima XR240amx system
with FlashPad HD detectors.

gehealthcare.com
 ′ = + √ + 4
2
1
Background
(3)

The mobile X-ray market is in transition as customers

move from traditional screen film based detectors
and CR detectors to wireless digital X-ray detectors Equation (1) and (3) show that the X-ray signal and
allowing near instantaneous image display, digital quantum noise are independent of pixel size; the
image processing and advanced applications. difference is in the pixel electronic noise appearing
Cassette size detectors for mobile systems are 14” by with a higher weight for the smaller pixel size
17” (35x43 cm) and 10” by 12” (25x30 cm). Data detector.
provided in this document is measured on 14” by 17”
detectors, performance of the 10” by 12” detector is The FlashPad HD electronic design improves the pixel
identical. noise performance reducing the electronic noise per
pixel versus FlashPad which makes lower than ,
moreover improvement on X-ray signal conversion is
Detector parameters comparison also realized allowing to convert more efficiently X-
ray signal into pixel intensity. Detective Quantum
In this paper, the Optima XR240amx system is Efficiency (DQE) is a metric considering all these
compared with the previous Optima XR220amx variables and is the most recognized and
system both equipped with a digital flat panel standardized method for detector performance
detector. The Optima XR240amx system introduces characterization [1], [2]. As part of Optima XR240amx
the FlashPad HD detector with a smaller pixel pitch of pilots installation, the lowest clinical detector
100µm while the pixel pitch of Optima XR220amx’s average entrance dose was observed for small
FlashPad is 200µm. pediatrics patient and was in the order of magnitude
of 1µGy, which is consistent with literature [3], [4]. As
Reduction of pixel size increased the Modulation shown on fig. 1 the DQE of FlashPad HD detector is
Transfer Function (MTF) thanks to a finer spatial higher than FlashPad detector at all dose levels. As
sampling of the signal. Additionally, the impact of per product data sheet, the high dose DQE at 0 lp/mm
pixel size on noise can theoretically be described as is 68% for FlashPad and 75% for FlashPad HD,
follows: considering the same input X-ray signal and a resulting in a 10% increase.
detector 1 with a pixel size of 200µm and a detector
2 with a pixel size of 100µm, the intensity of a 200µm
pixel is:

= ± √ +
2
1 (1)

With the X-ray signal, the quantum noise and

the pixel electronic noise. The intensity of a +10%
DQE

200µm pixel is:

FlashPad HD
= ± √ +
2
2 (2)
FlashPad

To display the same anatomical feature, detector 1

would use N1 pixels while detector 2 will use 4*N1
pixels. A fair comparison shall be to consider the

′ of 4 pixels of 100µm is:

signal per physical area and not per pixel. The signal
0.04 0.4 4 40
Dose [µGy]
′ =4∗ ± + + + + + + +
Fig. 1: DQE at 0lp/mm vs dose of FlashPad and FlashPad HD detectors

′ = + + + + + 4
The DQE improvement at low and high dose (fig. 1)
4 4 4 4 and low and high frequency (fig. 2) demonstrates that
the improvement in pixel design compensate the
theoretical noise increase to fully take advantage of

1
the higher MTF. As shown on fig. 2, the low frequency
performances of the two detector versions are
comparable. At high frequency, the efficiency of the
FlashPad HD detector is increased and allows the
detection of additional signal that is not detected by
FlashPad detector. Detector with 100µm pixel pitch
provides a Nyquist frequency at 5lp/mm (cutoff in
DQE curve in fig. 2) and has less artifact due to
aliasing as seen in fig. 3. Higher frequency noise
inherent to the signal is also detected giving a visual
difference as seen in fig. 3. Fig. 3: Resolution test pattern image, left Optima XR220amx, right
Optima XR240amx

Additionally, the bit depth is higher on Optima

XR240amx (16bits) than Optima XR220amx (14bits).

Detectability study with Contrast Detail

phantom

The Image Quality (IQ) is assessed in this section by

measuring the visibility of small and large targets of
low and high contrast under different clinically
relevant conditions. Such Contrast Detail (CD)
analysis is commonly used to assess Image Quality
and has good correlation with clinical image review by
radiologist for chest images [5]. The detection
Fig. 2: DQE at 0.35µGy vs frequency of FlashPad and FlashPad HD
capability of the Optima XR220amx system and
detectors Optima XR240amx system are compared using the
Artinis Medical Systems CDRAD 2.0 phantom. The
The detection of this high frequency signal can be phantom consists of a 265x265x10mm PMMA (Poly
easily illustrated by imaging a spatial resolution test (methyl methacrylate)) tablet with a matrix of 15
pattern which consists of a pattern of lead strips, of columns and 15 rows containing cylindrical holes with
different widths and distances in between (different variable diameter and depth. The diameter of the
spatial frequencies), sandwiched between plastic holes decreased from top to left and the depth of the
plates. Fig. 3 shows a resolution test pattern for which holes decrease from right to left. The less detectable
the frequency of 5lp/mm is visible on Optima holes are thus on the bottom left corner of the image
XR240amx system while Optima XR220amx system as can be seen in fig. 4. The top right region of the
is not able to detect this signal. Fine lead lines appear CDRAD phantom corresponds to large object with
blurred on the left image while they can be high contrast easily detectable, the top left region
distinguished on the right image. The spatial limiting corresponds to large object with low contrast and the
resolution is the highest spatial frequency detectable bottom right corresponds to small object with high
by the system. The theoretical spatial limiting contrast. The CD curve, delimiting visible and non-
resolution, linked to pixel size sampling, in vertical or visible objects, and the inverse of Image Quality
horizontal direction is 5lp/mm for Optima XR240amx Figure (IQF inv, index of image quality), computed
system and 2.5lp/mm for Optima XR220amx system. using equation (4), are used to assess Image Quality.
The limiting spatial resolution in diagonal direction, High detectability of a system corresponds to high
exceeding the resolution in vertical and horizontal IQF inv index.
direction, is measured with a resolution test pattern
at [5.6-6.3] lp/mm for Optima XR240amx system and
[2.5-2.8] lp/mm for Optima XR220amx system. The
resolution is increased by a factor of 2 between
Optima XR240amx and Optima XR240amx systems.

2
 The analysis software CDRAD Analyzer V2.1.15
developed by Artinis Medical Systems for their
CDRAD 2.0 phantom is used to automatically
compute CD curve and IQF inv. Two statistical values
can be adjusted by the user, the a-priori-difference-of
means (APD) and the significance level (α). The APD is
set relatively to the image depth, 4 for Optima
XR240amx and 1 for Optima XR220amx. A calibration
of CDRAD Analyzer software is performed to set the
significance level. The correlation of human observer
Higher Image Quality with software for review of CDRAD images is analyzed
by the review of 6 different images for each system
representing the diversity of setup and dose level
(second highest and second lowest dose level for
each condition) by 3 human reviewers. For each
image the human observer result is compared with
software result obtained with different significance
level alpha values to determine the significance level
Fig. 4: CDRAD phantom
alpha matching the human observer result as shown
on fig. 6.
(4)
Di = diameter of hole [mm] (Detail)
Ci = depth of hole [mm] (Contrast)

To represent a range of clinical applications the

CDRAD phantom was placed between additional
plates of PMMA following the three conditions shown
in fig. 5. Condition 1 with CDRAD phantom without Fig. 6: Schematic of the process used to determine the parameter of the
additional filtration represents thin anatomies CDRAD Analyzer software: the significance level alpha

(extremities, pediatrics), condition 2 with 10cm For Optima XR220amx system the significance level
PMMA filtration represents intermediate anatomies is set to 1E-2 and for Optima XR240amx it is set to 1E-
(C-spine, lower extremities) and condition 3 with 4 to fit the average of human observer review results
20cm PMMA represents thicker anatomies similarly to [6] as can be seen in fig. 7.
(Abdomen, Chest).

Fig. 5: Study conditions with CDRAD phantom

For condition 1 and 2, no grid is used and the source Fig. 7: Correlation between human observer (average) and software
to image distance (SID) is set to the detector gain CDRAD Analyzer V2.1.15 for different significance levels
calibration condition of 120cm. For condition 3, an
anti-scatter grid is used, 70lp/cm, 8:1 ratio and the Fig. 8 shows an example of CD curve with the same
SID is set to the grid focal distance of 130cm. For each significance level displaying the improvement in
mAs value 10 images are used for the CD curve and detectability for all hole diameters and depth. Using
IQF inv computation. The image processing the same significance level overestimate the
parameters used for each condition is system Factory difference between the two systems compared to
1. Hand PA protocol is used for condition 1, Thoracic- human review.
spine AP protocol is used for condition 2 and chest AP
protocol is used for condition 3.

3
 mAs
0 2 4 6 8 10

1.00

0.90
Improved detectability,
0.80
more holes are visible

Relative IQFinv
0.70

0.60

0.50 Optima XR240amx

0.40
Fig. 8: Example of Contrast Detail curve at 0.2mAs, condition 1 (no Optima XR220amx
PMMA), 60kV, SID 120cm, 10 images for computation with same 0.30
significance level
0.20
For each study condition, a wide range of tube 0 5 10 15 20 25 30 35
currents is used to represent a large range of clinical
Detector Dose [uGy]
applications, demonstrating the image quality
improvement at all relevant dose levels. Fig. 9, 10 Fig. 10: Relative IQFinv of Optima XR240amx and Optima XR220amx
and 11 show the IQFinv results for conditions 1, 2 systems for condition 2: 10cm additional PMMA filtration, 80kV, SID
120cm
and 3 respectively. For condition 1 the CDRAD
phantom is almost entirely visible with Optima For a clinically relevant exposure settings of a chest
XR220amx system as seen for image 5 and 6 in fig.7, X-ray in condition 3 (120kVp, 1mAs) the detectability
the CDRAD phantom shows limitation to measure improvement (IQFinv) measured is +40% as shown
the improvement of Optima XR240amx system in in fig. 11.
this condition.
mAs
mAs 0 2 4 6 8 10
0 1 2 3 4
1.00
1.00

0.95 0.80
Relative IQFinv

0.90
Relative IQFinv

0.60
0.85

0.80 0.40
up to +40%
Optima XR240amx
0.75
0.20 Optima XR240amx
0.70 Optima XR220amx
Optima XR220amx
0.65 0.00
0 5 10 15 20 25 0 1 2 3 4 5 6 7
Detector Dose [uGy] Detector Dose [uGy]

Fig. 9: Relative IQFinv of Optima XR240amx and Optima XR220amx Fig. 11: Relative IQFinv of Optima XR240amx and Optima XR220amx
systems for condition 1: no additional PMMA filtration, 60kV, SID 120cm systems for condition 3: 20cm additional PMMA filtration, 120kV, SID
130cm

4
The improvement is especially visible for small objects thanks to the resolution improvement. Fig. 12 shows the
example of the 2 last lines of the CDRAD images having the smallest hole diameters, 0.4 and 0.3 mm, acquired for
the two systems with condition 1 at 4mAs. To get comparable display conditions, the window center is set
proportionally to pixel intensity in the cell (1.3,0.3), and the window level is set proportionally to the noise. The third
image gives a schematic of the real position of the hole located in a corner for each cell, with the position the same
for all images. A black dot corresponds to a physical hole in the CDRAD phantom, and more holes are visible on
Optima XR240amx image.

Optima XR220amx

Optima XR240amx

Fig. 12: Example of CDRAD image for condition 1 with 4mAs, holes diameter is 0.4mm on the top line and 0.3mm on the bottom line
1st: Optima XR220amx; 2nd: Optima XR240amx
3rd: Schematic representation with real hole positions

5
Helix: optimized digital image processing

The Optima XR240amx with Helix image processing and FlashPad HD combine to provide exceptional resolution.
The Helix digital image processing has been optimized, especially to take advantage of the additional high frequency
signal contained in the raw image enhanced in the processed image. Extremities exams are a typical clinical
application which will use the additional anatomical small details to visualize trabecular structure for useful clinical
diagnostic. Fig. 13 shows a comparison of cadaver images of the wrist of the same patient imaged with Optima
XR220amx and Optima XR240amx systems. The overall sharpness of the image is visible on the standard size
image, and additional details are visible on the zoomed images. The improvements in image quality and visibility of
fine structures are especially visible in extremity images and higher resolution should also be beneficial for other
clinical applications.

Fig. 13: Cadaver wrist images of the same patient with identical exposure parameters: 60kV, 2.5mAs, SID 100cm
Left: Optima XR220amx system, top: standard view, bottom: zoom on the radial styloid process
Right: Optima XR240amx system, top: standard view, bottom: zoom on the radial styloid process
6
Exams assessing the positioning of small lines are another clinical application for which resolution and small details
detection are key parameters. A new custom image look for chest PA view has been developed taking advantages
of the flexibility in image processing parameters as well as the high resolution provided by the detector to increase
the visibility of small lines, the image can be easily reprocessed with this look in only one click with QuickEnhance.
Such images can be provided in addition to the standard image to help line placement assessment without
additional dose to the patient. Fig. 14 shows a real challenging clinical case acquired with an Optima XR240amx for
which the same X-ray acquisition is processed with standard settings and QuickEnhance, a new custom line
placement look. The nasogastric line pointed out by the arrow is not easily visible in the normal processed image
and is more clearly visible with QuickEnhance on the reprocessed image with a custom line placement processing
look.

Fig. 15 provides additional examples of clinical images of the one X-ray acquisition processed with two different
looks without additional dose to the patient.

Fig. 14: Challenging clinical case to assess nasogastric line placement acquired with Optima XR240amx system
Left: Normal image processed with standard image processing parameters, bottom: zoom on the NG line
Right: Reprocessed with QuickEnhance, a custom line look designed to enhance small lines visibility taking advantages of Optima XR240amx high resolution,
bottom: zoom on the NG line

7
 Fig. 15: Clinical cases to assess line placement acquired with Optima XR240amx system, top: adult case, bottom: pediatric case
Left: Normal image processed with standard image processing parameters
Right: Reprocessed with QuickEnhance, a custom line look designed to enhance small lines visibility taking advantages of Optima XR240amx high resolution

8
Conclusion Reference

The Optima XR240amx, with Helix image processing [1] Samei E, Murphy S, Christianson O (2013) DQE of
and FlashPad HD 100µm wireless cassette-size wireless digital detectors: comparative performance
digital flat panel detectors, delivers an increase of with differing filtration schemes. Med Phys.
resolution by a factor of 2 compared to Optima 40(8):081910
XR220amx. Additional high frequency signal is
detected and aliasing artifacts are reduced thanks to [2] IEC 62220-1-3 - Medical electrical equipment –
finer spatial sampling. This is achieved without noise Characteristics of digital X-ray imaging devices –
compromise thanks to an improved pixel electronic Part 1-3: Determination of the detective quantum
design allowing a 10% increase of the low frequency efficiency
DQE compared to Optima XR220amx. Helix digital [3] Hamer, O. W. et al. (2005) Chest Radiography
image processing is optimized to take advantage of with a Flat-Panel Detector: Image Quality with Dose
this additional signal and provide sharper processed
Reduction after Copper Filtration. Radiology 237,
images. Overall up to 40% detectability increased is 691–700
measured with a Contrast Detail phantom study
compared to Optima XR220amx. [4] Knight S P (2014) A paediatric X-ray exposure
chart. Journal of Medical Radiation Science 61(3):
191–201

[5] De Crop A, Bacher K, Van Hoof T, et al. (2012)

Correlation of contrast-detail analysis and clinical
image quality assessment in chest radiography with
a human cadaver study. Radiology 262, 298–304.

[6] Pascoal A, Lawinski C P, Honey I et al. (2005)

Evaluation of a software package for automated
About GE Healthcare quality assessment of contrast detail images –
comparison with subjective visual assessment. Phys
GE Healthcare provides transformational medical
technologies and services to meet the demand for Med Biol 50:5743-5757
increased access, enhanced quality, and more
affordable healthcare around the world. GE works on
things that matter – great people and technologies
taking on tough challenges. From medical imaging,
software & IT, patient monitoring and diagnostics to GE Healthcare
drug discovery, biopharmaceutical manufacturing 3000 North Grandview
technologies and performance improvement Waukesha, WI 53188
solutions, GE Healthcare helps medical professionals USA
deliver great healthcare to their patients. www.gehealthcare.com

imagination at work
9