Image quality assessment in digital mammography: part I. Technical characterization of the
systems
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
2011 Phys. Med. Biol. 56 4201
(http://iopscience.iop.org/0031-9155/56/14/002)
Download details:
IP Address: 155.105.7.44
The article was downloaded on 14/09/2011 at 09:07
Please note that terms and conditions apply.
View the table of contents for this issue, or go to the journal homepage for more
Home Search Collections Journals About Contact us My IOPscience
IOP PUBLISHING PHYSICS IN MEDICINE AND BIOLOGY
Phys. Med. Biol. 56 (2011) 4201–4220 doi:10.1088/0031-9155/56/14/002
Image quality assessment in digital mammography:
part I. Technical characterization of the systems
N W Marshall
1
, P Monnin
2,3
, H Bosmans
1
, F O Bochud
2
and
F R Verdun
2
1
UZ Gasthuisberg, Department of Radiology, Herestraat 49, 3000 Leuven, Belgium
2
University Institute for Radiation Physics (IRA), CHUV, UNIL, Grand Pr
´
e 1, 1007 Lausanne,
Switzerland
3
Haute Ecole Cantonale Vaudoise de la Sant
´
e (HECVSant
´
e), Fili
`
ere TRM,
Avenue de Beaumont 21, 1011 Lausanne, Switzerland
E-mail: nicholas.marshall@uz.kuleuven.ac.be
Received 1 March 2011, in final form 4 May 2011
Published 23 June 2011
Online at stacks.iop.org/PMB/56/4201
Abstract
In many European countries, image quality for digital x-ray systems used
in screening mammography is currently specified using a threshold-detail
detectability method. This is a two-part study that proposes an alternative
method based on calculated detectability for a model observer: the first
part of the work presents a characterization of the systems. Eleven digital
mammography systems were included in the study; four computed radiography
(CR) systems, and a group of seven digital radiography (DR) detectors,
composed of three amorphous selenium-based detectors, three caesium iodide
scintillator systems and a silicon wafer-based photon counting system. The
technical parameters assessed included the system response curve, detector
uniformity error, pre-sampling modulation transfer function (MTF), normalized
noise power spectrum (NNPS) and detective quantum efficiency (DQE).
Approximate quantum noise limited exposure range was examined using a
separation of noise sources based upon standard deviation. Noise separation
showed that electronic noise was the dominant noise at low detector air
kerma for three systems; the remaining systems showed quantum noise limited
behaviour between 12.5 and 380 μGy. Greater variation in detector MTF was
found for the DR group compared to the CR systems; MTF at 5 mm
−1
varied
from 0.08 to 0.23 for the CR detectors against a range of 0.16–0.64 for the DR
units. The needle CR detector had a higher MTF, lower NNPS and higher DQE
at 5 mm
−1
than the powder CR phosphors. DQE at 5 mm
−1
ranged from 0.02
to 0.20 for the CR systems, while DQE at 5 mm
−1
for the DR group ranged
from 0.04 to 0.41, indicating higher DQE for the DR detectors and needle CR
system than for the powder CR phosphor systems. The technical evaluation
0031-9155/11/144201+20$33.00 © 2011 Institute of Physics and Engineering in Medicine Printed in the UK 4201
4202 N W Marshall et al
section of the study showed that the digital mammography systems were well
set up and exhibiting typical performance for the detector technology employed
in the respective systems.
1. Introduction
Before a digital mammography x-ray system can be used for breast screening in many European
countries, the system must meet the minimum image quality performance defined in the
current edition of the European Guidelines for Quality Assurance in Mammography Screening
(European Commission 2006), within the dose limits given in this document. While not an
official document that is cast directly in European legislation, these guidelines are extremely
influential and have been adopted as de facto minimum performance standards in many
European countries (see for example NHSBSP (2009a)).
The image quality standard is specified in terms of threshold-detail detectability for a
range of circular discs. The detectability test as presented in the guidelines is used as a
system test, rather than a test that focuses solely on x-ray detector performance, as is the
case for fluoroscopy detectors in the UK (Hay et al 1985). A range of system factors
will therefore influence the ability of a unit to meet the standard in the guidelines; these
include x-ray tube potential, tube anode and filter combination, scatter rejection method and
imaging performance of the x-ray detector, often specified via the detective quantum efficiency
(DQE). Although presented as a system test, threshold contrast-detail detectability test objects
often have a limited dynamic range and generate images that are somewhat removed from
what can be considered typical patient content in terms of greyscale, spatial frequency and
contrast range. Further limitations of this method are examined in the second part of this
study.
Along with the choice of operating point for the automatic exposure control (AEC),
one of the main parameters determining system imaging performance is the x-ray detector
(Aufrichtig 1999, Samei and Flynn 2003). Detector performance can be assessed using
quantitative measurements such as the modulation transfer function (MTF), normalized noise
power spectrum (NNPS) and DQE (Metz et al 1995, Samei et al 2006, Dobbins et al 2006).
While not the final determinant of system image quality, these parameters have considerable
impact on the quality of images produced by an imaging system (Aufrichtig 1999, Marshall
2006a). Part two of this work proposes a model observer approach for use in screening
mammography image quality specification and correlates the calculated detectability index
with the standard metric, that of threshold-detail detectability measured using the CDMAM
test object. Given that these are system specific rather than detector specific metrics of image
quality, it is important to establish the performance of the detectors using standard quantitative
metrics. This will help to isolate elements of system performance related to the intrinsic
detector quality from those related to the AEC operating point, x-ray contrast and scatter
rejection methods. Furthermore, the first part of the study provides a useful comparison of
a range of currently available mammography x-ray detectors. The aims of the paper were
therefore to characterize technical performance using quantitative measures such as response
curve, detector uniformity error, MTF, NNPS and DQE for the 11 x-ray detectors in the
study. Finally, data presented in this part should confirm whether the performance for a given
detector could be considered typical when evaluated by the threshold-detail detectability
method and the calculated detectability index, as described in the second part of this
work.
Image quality assessment in digital mammography 4203
Tab le 1. Characteristics of the x-ray detectors assessed in this study along with the detector
response type and fit coefficients.
Pixel Detector
Detector pitch Pixel response
name Technology (μm) matrix curve A B
Agfa MM 3.0R Single-side powder CR 50 4708 × 5844 Power 695.73 0.521
Agfa HM 5.0 Single-side needle CR 50 4708 × 5844 Power 702.53 0.518
Fuji Profect Dual-side powder CR 50 3540 × 4740 Logarithmic −570.46 228.18
Carestream Single-side powder CR 48.5 3584 × 4784 Logarithmic 2858.5 −359.37
EHR-M3
Fuji Amulet a-Se
/optical switch 50 3540 × 4740 Logarithmic −1010.3 1874.4
GE Senographe CsI
/a-Si TFT switch 100 1914 × 2294 Linear −3.91 10.08
2000D
GE Senographe DS CsI
/a-Si TFT switch 100 1914 × 2294 Linear −12.57 8.83
GE Essential CsI
/a-Si TFT switch 100 2394 × 3062 Linear −12.65 7.75
Hologic Selenia a-Se
/TFT switch 70 2140 × 2140 Linear 41.21 4.08
Sectra MDM Photon counter
/ 50 4915 × 5355 Linear – –
Si-wafer
Siemens Inspiration a-Se
/ TFT switch 85 2658 × 3318 Linear 50.31 3.24
2. Materials and methods
2.1. Digital mammography detectors studied
Eleven digital mammography systems were included in this study, with examples of all the
commercially available detector technologies. Four computed radiography (CR) systems were
assessed, including two single-sided readout powder phosphor systems, a dual-sided readout
units used with a powder phosphor and a single-sided readout system used with a needle
CR phosphor. Of the remaining systems, six used flat-panel detectors. Three of these were
indirect conversion units that used a caesium iodide phosphor bonded to a light-sensitive thin
film transistor (TFT) array formed from amorphous silicon (a-Si). Two further flat-panel
systems were evaluated; these were direct conversion detectors using amorphous selenium
(a-Se) in conjunction with a TFT readout array. The final unit utilized a silicon wafer photon
counter with pre- and post-breast slit collimation. Basic technical parameters for the systems
are given in table 1.
CR detectors are generally not integrated into a given x-ray mammography system but
can be used with x-ray units from different suppliers. Both Agfa CR detectors in this study
were used with a Siemens Mammomat 3000 system while the Fuji Profect cassettes were used
with a Siemens Mammomat 3000 Nova x-ray system. Images with the Carestream EHR-M3
were acquired with a Trex Benett Contour 2000 unit. CR system performance, in terms of
image quality produced for a given mean glandular dose (MGD), therefore depends on the
anode/filter (A/F) settings available on x-ray unit with which the cassettes are used. Agfa give
explicit A/F recommendations for different x-ray systems while Fuji make no recommendation
for the Profect CR system and hence A/F used will often be chosen in conjunction with the
local Medical Physics department. The Trex system, used for the Carestream EHR-M3
CR acquisitions, only had molybdenum/molybdenum (Mo/Mo) A/F available. There is no
common A/F setting available between the 11 units and hence a common energy could not be
selected for the objective image quality evaluations.
4204 N W Marshall et al
Tab le 2. Acquisition factors used for the evaluation along with number of photons mm
−2
μGy
−1
;
approximate air kerma at the detector for the AEC mode; air kerma at the detector for the DQE
evaluation; detector uniformity error (coefficient of variation (%)).
Air kerma K for DQE Detector
Tube potential
/ at detector and NNPS uniformity
Detector anode
/ q
0
for AEC evaluation error
name filter (mm
−2
μGy
−1
) mode (μGy) (μGy) (%)
Agfa MM 3.0R 29 kV Mo/Rh 5662 102 115 13
Agfa HM 5.0 29 kV Mo
/Rh 5662 104 115 14
Fuji Profect 27 kV Mo
/Rh 5462 64 124 9.1
Carestream EHR-M3 28 kV Mo
/Mo 5057 65 92 17
Fuji Amulet 29 kV W
/Rh 6168 88 96 1.4
GE Senographe 2000D 28 kV Rh
/Rh 6118 66 91 2.7
GE Senographe DS 28 kV Mo
/Mo 5057 98 86 2.4
GE Essential 29 kV Rh
/Rh 6248 83 92 2.1
Hologic Selenia 29 kV W
/Rh 6168 100 104 2.7
Sectra MDM 28 kV W
/Al 6249 64 104 –
Siemens Inspiration 28 kV W
/Rh 6070 98 102 1.2
2.2. Test equipment
First, a note on the equipment used for the evaluations. As image data were acquired from
x-ray machines located at two locations (Switzerland and Belgium), two sets of test equipment
had to be used to acquire these data (two different 2 mm Al filters, dosemeters and MTF
edges). The dosemeters were a Radcal monitor (Radcal, Monrovia, USA) used with a
6cm
3
ionization chamber (10×5–6 M) and an RTI Barracuda (M
¨
olndal, Sweden) used with a
solid-state multipurpose detector (MPD). Calibration of both devices was traceable to national
standards. As an additional check on consistency of the output measurements, the two
dosemeters were brought together for comparative measurements on a single mammography
system. Output with 2 mm added Al in the x-ray beam was measured as a function of tube
current-time product (mAs) with the two dosemeters. Two MTF edges were used; a Tungsten
square of side 50 mm and thickness 0.5 mm and a steel rectangle of dimensions 60 mm ×
120 mm and thickness of 1 mm.
2.3. Detector response
The first step in the technical characterization was the measurement of the detector response
at the tube potential and A/F given in table 2, with a 2 mm Al filter of 99% purity placed at
the x-ray tube port. Air kerma was measured at the breast support platform as a function of
tube mAs and corrected by the inverse square law to give the air kerma at the detector (K);
no correction was made for the transmission of detector covers. The detector response was
then measured from uniform exposure (flood) images acquired as a function of air kerma at
the detector. All detectors were fully irradiated (open collimation) for the flood acquisitions
except for the Sectra MDM where a 12.8 cm × 12.8 cm collimated field was used. For the
systems with integrated flat-panel detectors the antiscatter grid was removed, while for CR
systems the cassette was placed on the breast support platform. The target air kerma values
were 12.5, 25, 50, 100, 200 and 400 μGy at the detector, although this range could not be set
for all systems. ‘For Processing’ DICOM images were then acquired; with the integrated flat-
panel units, standard corrections for x-ray heel effect and detector offset, gain and defective
