Eur Radiol (2007) 17: 2934–2940
DOI 10.1007/s00330-007-0650-x
CHEST
Claus Koelblinger
Mathias Prokop
Michael Weber
Johannes Sailer
Fabiola Cartes-Zumelzu
Cornelia Schaefer-Prokop
Received: 29 May 2006
Revised: 8 March 2007
Accepted: 30 March 2007
Published online: 10 May 2007
# Springer-Verlag 2007
Two K versus 4 K storage phosphor chest
radiography: detection performance
and image quality
Abstract The purpose of this study
was to evaluate the effect of matrix
size (4-K versus 2-K) in digital storage
phosphor chest radiographs on image
quality and on the detection of CT-
proven thoracic abnormalities. In 85
patients who underwent a CT of the
thorax, we obtained two additional
posteroanterior storage phosphor chest
radiographs, one with a matrix size of
3,520×4,280 (=4-K) and the other with
a matrix size of 1,760×2,140 (=2-K).
Acquisition, processing and presenta-
tion parameters were identical for all
radiographs. Two radiologists evalu-
ated the presence of mediastinal,
pleural, and pulmonary abnormalities
on hard copies of the radiographs,
applying ROC analysis. In addition,
four radiologists were asked to
subjectively rank differences in
image quality and to assess the
demarcation of anatomic landmarks
comparing the images in a blinded
side-by-side manner. These data
were analyzed using a two-sided
binomial test with a significance
level of P<0.05. Both tests, the
ROC analysis of the detection per-
formance and the binomial test of
the subjective quality ratings, did
not reveal significant differences
between the two matrix sizes.
Compared to 2-K radiographs, 4-K
storage phosphor chest radiographs
do not provide superior detection
performance or image quality when
evaluated in identical hard copy
formats.
Keywords Digital radiography
.
Chest radiography
.
Spatial
resolution
.
Image quality
.
ROC analysis
Introduction
The necessary spatial resolution capacity for digital chest
radiography has been an intensely discussed topic since the
introduction of digital imaging systems in the early 1980s.
For a long time, the smallest available pixel size for digital
chest radiographic imaging systems was 0.2 mm, with a
matrix size of 2 K (1,760×2,140 pixel). This matrix size
provided a maximum spatial resolution of 2.5 lp/mm
compared to 8 lp/mm of conventional film systems when
measured with a lead bar phantom [1]. Results of previous
clinical as well as phantom studies are controversial with
regard to whether the lower spatial resolution would lead to a
loss of diagnostic information compared to conventional film.
Some studies reported that the lower spatial resolution would
result in a lower detectability of linear or micro-nodular
opacities, as seen in pleural lines or interstitial disease [2–4].
Other authors, however, concluded that 2-K hardcopy chest
radiographs perform equivalently to film if adequate process-
ing is applied [5–8, 10]. Similarly it has been shown in
contrast detail phantom studies that detail detectability is
described insufficiently by pixel size and spatial resolution
alone, but relies on a complex interaction of contrast and
spatial resolution more adequately described by the modula-
tion transfer function (MTF; [9, 11, 16]).
Nowadays, most manufacturers also offer storage
phosphor systems with a pixel size of 0.1 mm for the
large 35×43 cm detector format required for the chest,
C. Koelblinger (*)
.
M. Weber
.
J. Sailer
.
F. Cartes-Zumelzu
Department of Radiology,
Medical University of Vienna,
Waehringer Guertel 18-20,
A-1090 Vienna, Austria
e-mail: claus.koelblinger@meduniwien.ac.at
Tel.: +43-1-404004891
Fax: +43-1-404004898
M. Prokop
Department of Radiology,
University Medical Center Utrecht,
Utrecht, The Netherlands
C. Schaefer-Prokop
Department of Radiology,
Academic Medical Center (AMC),
University of Amsterdam,
Amsterdam, The Netherlands