Download Comparison of clinical and physical measures of image quality in

Institutionen för medicin och vård
Avdelningen för radiofysik
Hälsouniversitetet
Comparison of clinical and physical measures
of image quality in chest PA and pelvis AP
views at varying tube voltages
Gustav Ullman, Michael Sandborg, Anders Tingberg,
David R Dance, Roger Hunt and Gudrun Alm Carlsson
Department of Medicine and Care
Radio Physics
Faculty of Health Sciences
Series: Report / Institutionen för radiologi, Universitetet i Linköping; 98
ISRN: LIU-RAD-R-098
Publishing year: 2004
© The Author(s)
Report 98
Dec. 2004
ISRN ULI-RAD-R--98--SE
Comparison of clinical and physical measures of
image quality in chest PA and pelvis AP views at
varying tube voltages
G Ullman1, M Sandborg1, Anders Tingberg2, D R Dance3,
Roger Hunt3 and G Alm Carlsson1
1
2
3
Department of Radiation Physics, Linköping University
Department of Radiation Physics, Malmö University Hospital
Department of Physics, The Royal Marsden NHS Trust, London
Full addresses:
1 Department of Radiation Physics, IMV,
Faculty of Health Sciences,
Linköping University,
SE-581 85 LINKÖPING,
Sweden
2 Department of Radiation Physics,
Malmö University Hospital
SE-20502 MALMÖ
Sweden
3 Joint Department of Physics,
The Royal Marsden NHS Trust and
Institute of Cancer Research,
Fulham Road,
London SW3 6JJ,
United Kingdom
-1-
Abstract
Image quality in digital chest PA and pelvis AP was assessed using two different methods; one
based on observations of images of an anthropomorphic phantom, one based on computer
modelling using an anthropomorphic voxel phantom. The tube voltage was varied within a
broad range (50-150 kV), including those values typically used with screen-film radiography.
The tube charge was altered so that approximately the same effective dose was achieved in
the modelled patient (anthropomorphic phantom). Two x-ray units were employed using a
digital image detector (computed radiography, CR, system) with standard tube filtration and
anti-scatter device. Clinical image quality was assessed by a group of radiologists using a
visual grading analysis (VGA) technique based on the revised CEC image criteria. Physical
image quality was derived from the computer model in terms of the signal-to-noise ratio, SNR
for fixed effective dose in the voxel phantom. The computer model uses Monte Carlo
simulations of the patient and complete imaging system. Both the VGAS (visual grading
analysis score) and SNR increase with decreasing tube voltage in both chest PA and pelvis AP
examinations, indicating superior performance if lower tube voltages than used today are
employed in digital radiology. A positive correlation between clinical and physical measures of
image quality was found. The pros and cons of using lower tube voltages with CR digital
radiography than typically used in analogue screen-film radiography are discussed as well as
the relevance of using VGAS and quantum noise SNR as measures of image quality.
Key words: image quality, effective dose, tube voltage, visual grading analysis,
signal-to-noise ratio, chest radiography, pelvis radiography, computed radiography
-2-
Introduction
Since the use of digital techniques in radiographic imaging is rapidly growing, it is important to
assess and review the techniques used to obtain the images. The tube voltage is one of the
variables that can be readily altered prior to exposure of each patient and view. The selection
of appropriate tube voltage in analogue, screen-film radiology is a balance between the
appropriate image quality in all relevant parts of the image and of patient dose. In digital
radiography, the contrast can be manipulated in the display of the image and the patient dose
is not determined by properly exposing the film but by achieving an acceptable noise-level in
the image.
Here, optimisation means searching for the tube voltage that maximises the quantities
corresponding to image quality, and at the same time maintaining an acceptable dose level to
the patient. Methods to assess the patient dose are readily available whereas methods to
assess clinical image quality are still developing. Image quality must always be related to the
clinical questions at hand. As these may vary it has been argued that image criteria (CEC
1996), that describe features that characterise good clinical images, may be used to assess the
over-all quality of the image. By allowing a group of experienced radiologists to give a graded
response of the quality of the structures that the criteria refer to in images obtained with
different tube voltages, it may indicate which tube voltage settings that would be preferable.
An alternative method is to assess the quality of the image data itself by using the signal-tonoise ratio, SNR, of specified simulated structures as a measure of image quality. This relies on
the fact that previous studies have found a positive correlation between SNR of particular
structures and the assessment of the image criteria (Sandborg et al 2001, Tingberg et al
2004). In order to make the assessment more realistic, an anthropomorphic phantom and a
computer model of the imaging system was used to derive the SNR of structures
corresponding to the CEC image criteria.
In Tingberg and Sjöström (2003) chest and pelvis images of anthropomorphic phantoms were
assessed by a group, of radiologists using visual grading analysis, VGA. They conclude that, for
the same effective dose, higher VGA-score (VGAS) was found using lower tube voltages in
both chest PA and pelvis AP views. Similar trends were reported by other studies (Dobbins et
al 1992, Chotas et al 1993, Sandborg et al 1994 and Launders et al 2001). The aim of this
work is to explore if the present Monte Carlo model predicts the same dependence of image
quality on tube voltage as Tingberg and Sjöström (2003).
Materials and method
Computer model of chest and pelvis examination
Patient model
The Monte Carlo computer program uses a voxel phantom developed by Zubal et al (1994) to
simulate the transport of x-ray photons through the patient. The voxel phantom is segmented
into several different organs. Five types are used in the model: soft tissue, bone, bone
marrow, lung tissue and air. The individual voxels are approximately 3x3x4 mm3. The anteriorposterior length is 24 cm across the thorax, corresponding to an average-sized patient.
The program computes quantities of two main types: quantities associated with image quality
and with patient dose.
Imaging system model
A description of the Monte Carlo photon transport code can be found in McVey et al (2003) and
is briefly summarised here. The main components of the imaging system are included in the
simulation, i.e. the x-ray spectrum (tube voltage and total filtration), patient and patient couch
or chest stand, anti-scatter grid, cassette front and image detector. The x-ray spectra were
obtained from Birch et al (1979). Data on thickness of the couch top and chest stand were
derived from measurements. The manufacturer provided data on the construction of the antiscatter grid. A computed radiography (CR) system was simulated with a surface-weight of the
BaFCl-phosphor active layer of 100 mg/cm2.
-3-
Dosimetric quantities
The energy imparted per unit area to the image detector was normalised to a fixed value of
patient effective dose for each examination: chest 20 µSv and pelvis 150 µSv. The effective
dose was computed according to ICRP 90 (1991). Equivalent doses to relevant organs were
obtained by dividing the energy imparted to an organ of interest with the mass of that organ.
The organ equivalent doses were then multiplied with the appropriate tissue-weighting factor
and summed to form the effective dose.
Imaging system parameters
Table 1 specifies the imaging system parameters of two x-ray units used in the experiments.
Two anthropomorphic phantoms were imaged to mimic a chest PA and a pelvis AP
examination, respectively (Tingberg and Sjöström 2003). Ten different tube voltages were
used including both tube voltages lower and higher than those used at a typical patient
examination with analogue screen-film system. The tube charge was modified in order to
achieve the same effective dose for each projection and for all selected tube voltages. This was
accomplished by measuring the kerma-area-product and using conversion factors between the
kerma-area-product and effective dose from Hart et al (1994). The standard tube voltage used
for chest PA and pelvis AP examinations in the clinic were 125 kV and 70 kV, respectively.
These tube voltages are here denoted the reference tube voltages, Uref.
Table 1. The basic imaging system parameters used with the two x-ray units used in this
study.
Tube voltage, U
Reference tube voltage, Uref
Filtration
Focus detector distance, FDD
Field area
Grid ratio, r
Strip density, N
Detector name
Detector material
Surface density
Chest PA
Pelvis AP
70-150 kV
50-102 kV
125 kV
70 kV
7 mm Al
6 mm Al
155 cm
120
32.8 x 22.8 cm2
32.0 x 28.0 cm2
12
12
36 cm-1
36 cm-1
Fuji FCR AC-3, ST V plates Fuji FCR AC-3, ST V plates
BaFCl
BaFCl
100 mg/cm2
100 mg/cm2
Assessment of image quality
Clinical image quality
The ten images of each view were evaluated by five (chest) and six (pelvis) experienced
radiologists according to slightly modified (Tingberg et al 2000, Sund et al 2000) CEC image
criteria (CEC 1996). The criteria are described in table 2. The radiologists were asked to give a
graded response of the quality of the imaged structures mentioned in the criteria. Two images
were presented simultaneously and the image to the right was assessed as either being clearly
inferior (VGA=-2), inferior (-1), equal to (0), superior (+1) or clearly superior (+2). The VGA
scores (VGAS) were averaged over all imaged structures and all radiologists to form an
average VGAS for each tube voltage.
The average values of the VGAS derived as described above are given in figure 1 (Tingberg
and Sjöström 2003). These figure show that the VGAS decreases with increasing tube voltage
at a rate that is larger for the pelvis examination.
-4-
Table 2. Structures used in the VGA evaluation (Tingberg et al 2003).
1
2
3
4
5
6
Chest PA
Vessels seen 3 cm from the pleural margin
Thoracic vertebra behind the heart
Retro-cardiac vessels
Pleural margin
Vessels seen an face in the central area
Hilar region
Pelvis AP
Sacrum (spongiosa)
Sacral foramina
Pubic and ishial rami
Sacroiliac joints
Femoral bilateral
Pelvis AP
2
2
1,5
1,5
1
1
0,5
0,5
0
-0,5
0
20
40
60
80 100 120 140
VGAS
VGAS
Chest PA
0
-0,5
-1
-1
-1,5
-1,5
-2
-2
0
20
40
60
80
100 120 140
Tube voltage (kV)
Tube voltage (kV)
(a)
(b)
Figure 1. The visual grading analysis score (VGAS) for (a) the chest and (b) for the pelvis
images as function of the tube voltage. The VGAS values represent averages over all imaged
structures (table 2) and radiologists. The uncertainty bars indicate one standard error. The
solid line (r2=0.90) indicates that there is a linear relationship between VGAS and tube voltage
(data from Tingberg and Sjöström 2003).
Physical image quality
Physical image quality was computed for a set of details included in the voxel phantom at
specified locations according to figure 2. Table 3 gives the detail sizes and compositions used
in the calculations. The details and their positions were chosen so as to correspond to the
structures in table 2.
Table 3. Structure details used in the model calculations. The compositions and densities of
bone, blood, soft and lung tissues are obtained from ICRU 46.
1
2
3
4
5
6
Chest PA
2 mm blood in lung tissue
1 mm bone in soft tissue
2 mm blood in lung tissue
1 mm soft tissue in lung tissue
2 mm blood in lung tissue
2 mm blood in lung tissue
Pelvis AP
1 mm bone
1 mm bone
1 mm bone
1 mm bone
1 mm bone
-
in
in
in
in
in
soft
soft
soft
soft
soft
tissue
tissue
tissue
tissue
tissue
The ideal observer signal-to-noise ratio SNR (ICRU 1996) was used as the measure of physical
image quality. The computer program calculates this SNR for a detail with the area of a pixel,
here called the SNRM
-5-
SNRM =
N p 1 ⋅ ε ′p 1 − N p 2 ⋅ ε ′p 2
(1)
2
2
⋅ ε ′p 1 + N s ⋅ ε s′
N p1
where N is the number of photons incident on each pixel. The indices p and s represent
contributions from primary and scattered photons, respectively. The index n=1 refers to a
pixel behind the detail, and n=2 refers to the same pixel with the detail absent. The quantities
ε´ and ε´2 are the mean and mean squared values of the energy imparted per incident photon
for the specific pixel. The operator <…> denotes the statistical expectation (mean) value of the
specified quantity. From the SNRM it is possible to get the SNR for a given detail as
SNR 2 = SNRM2 ⋅
A 2
⋅ rDF
ap
(2)
where A is the projection area of the detail, ap is the pixel area and r2DF is the signal to noise
ratio degradation factor. The latter arises from imaging system unsharpness and additional
detector noise and is derived separately according to Sandborg et al (2003).
For each contrasting detail, n, in table 3 (corresponding to the structures in table 2), the SNRn
relative to its value at the reference tube voltage, SNRn(Uref), was computed (SNRn(U)/
SNRn(Uref)). These ratios were then averaged for all the structures and a quantity
δSNR (U ) computed as given in equation 3. Here, unity was subtracted from the average value
in order to obtain the value zero for the reference tube voltage and allow negative values when
the image quality is inferior to that of the reference system (corresponding to the scale of VGA
values).
δSNR (U ) =
(a)
1
N
SNRn (U )
∑ SNR (U ) − 1
n
n
(3)
ref
(b)
Figure 2. The two computed images (a) chest PA and (b) pelvis AP show the locations of the
details (indicated by red solid rings) included in the voxel phantom which were used in the
simulations to correspond to the structures in table 2.
-6-
Results
Figures 3 show the SNR of the selected details as function of tube voltage for the chest PA and
pelvis AP examination. The SNR for fixed effective dose decreases with increasing tube voltage
for both chest PA and pelvis AP views and for all details. The rate of descent is more rapid for
the pelvis examination.
Pelvis AP
30
60
25
50
20
40
SNR
SNR
Chest PA
15
30
10
20
5
10
0
0
20
40
60
0
80 100 120 140
0
20
Tube Voltage (kV)
40
60
80 100 120 140
Tube Voltage (kV)
(a)
(b)
Figure 3. The SNR of the details in table 3 as function of tube voltage for (a) chest PA and (b)
pelvis AP. The effective dose is constant.
Figure 4 shows the average relative change in SNR, δSNR , with increasing tube voltage. In
the chest PA, the δSNR is 37% higher at 70kV and 12% lower at 150 kV compared to the
reference tube voltage at 125 kV. The corresponding values of δSNR for the pelvis AP
examination are 37% higher and 44% lower compared to 70 kV.
Pelvis AP
0,5
0,5
0,4
0,4
0,3
0,3
0,2
0,2
0,1
0,1
SNR
SNR
Chest PA
0
-0,1 0
20
40
60
80 100 120 140
0
-0,1 0
-0,2
-0,2
-0,3
-0,3
-0,4
-0,4
20
40
60
80 100 120 140
-0,5
-0,5
Tube Voltage (kV)
Tube Voltage (kV)
(a)
(b)
Figure 4. The average relative change in SNR, δSNR , in (a) a chest PA and in (b) a pelvis AP
examination as function of tube voltage. The effective dose is constant.
-7-
Figure 5 shows the relation between clinical image quality measured using visual grading
analysis score (VGAS) and the physical measure of image quality, quantified by the relative
change in SNR, δSNR . There is a positive linear relationship between the two measures of
image quality indicating that the SNR is related to the radiologists’ grading of the image
criteria.
Pelvis AP
-0,5
-0,3
2
2
1
1
0
-0,1
0,1
0,3
0,5
VGAS
VGAS
Chest PA
-0,5
-0,3
0
-0,1
-1
-1
-2
-2
δSNR
0,1
0,3
0,5
δSNR
(a)
(b)
Figure 5. The relation between VGAS and δSNR for a simulated (a) chest PA and (b) pelvis AP
examination. The error bars correspond to one standard error in the results of the observation
studies (fig 1). The r2 of the fitted line (chest PA: r2=0.91, pelvis AP: r2=0.94) indicate that the
VGAS and δSNR are linearly correlated.
Discussion
The results in Tingberg and Sjöström (2003) indicate that the optimal tube voltage in chest PA
and pelvis AP examinations with digital image detectors (computed radiography, CR) is lower
than the tube voltages typically used with analogue, screen-film systems (125 kV chest and
70 kV pelvis). This is concluded since both VGAS in their study and the SNR found in the
present study increases with decreasing tube voltage when the same effective doses are used
for all tube voltages.
A number of reasons for using lower tube voltages with digital detectors compared to screenfilm systems are listed by Tingberg and Sjöström (2003). (1) The image detector has a larger
absorption efficiency of x-ray photons at low photon energies. The detective quantum
efficiency (DQE) typically increases with decreasing tube voltage since the quantum absorption
efficiency of the active detector material increases with decreasing photon energy. Also the Kedge of barium in BaFCl (used in CR systems) is 37 keV that is lower than the corresponding
value for gadolinium in Gd2O2S (often used in screen-film systems) at 50 keV. The lower Kedge value results in inferior absorption of the high-energy x-ray photons present at higher
tube voltages. (2) The object contrast, and hence also the signal in the signal-to-noise ratio
expression, decreases with increasing tube voltage. (3) The contribution to the effective dose
per entrance air kerma or kerma-area-product is lower for low tube voltages compared to high
tube voltages. Lower tube voltages therefore results in lower effective dose if the entrance air
kerma at the patient is similar. (4) Additional arguments for using lower tube voltages are
found in Ullman et al (2004). The SNR2/E is between 40-70% higher (depending on chest
region) at 90 kV compared to 150 kV (Ullman et al (2004) for the chest examination,
indicating that for the same effective dose the SNR increases with decreasing tube voltage.
-8-
There are, on the contrary also a number of arguments for keeping the tube voltages at the
same comparably high values as used with screen-film radiography. The arguments are: (1)
The object contrast of soft tissues (vessels, metastasis) in the lung relative to that of bony
structures (ribs, thoracic spine) will be higher using higher tube voltages (Ullman et al 2004).
Hence the appearance of anatomical background structures such as the ribs will be enhanced if
the tube voltage is reduced that may impede the radiologist to detect pathological lesions of
interest. (2) The tube charge (and exposure time) will be larger when low tube voltages are
used due to the reduced radiation yield and higher attenuation in the patient compared to
using higher tube voltages. This may, in particular, be an issue in chest imaging if the
increased exposure time results in severe motion unsharpness. Also the geometrical (focal
spot) unsharpness may deteriorate since the focal spot size typically increases with decreasing
tube voltage and increasing tube charge. (3) The settings of the automatic exposure control
(AEC) ion-chambers, that terminate the exposure when a predefined air kerma has been
achieved, may need to be revised. Ideally they should terminate the exposure when the
desired noise-level (SNR per pixel) is achieved and not at a value corresponding to a set target
optical density on the film. (4) The difference in the energy imparted to the detector per unit
area beside and just behind the contrasting detail (signal-difference or object contrast) divided
by the difference between the maximum (95% percentile) and minimum (5% percentile)
energy imparted to the detector per unit area in the whole image (image dynamic range), is
approximately constant or increases slowly with increasing tube voltage in chest PA imaging
(Ullman et al. 2004). This indicates that the higher tube voltages are preferable.
By reducing the tube voltage one may not comply with recommendations by national (SSI
2002) and international (CEC 1996) bodies on what tube voltage to use for these examination.
The Swedish recommendations SSI (2002) suggest use of 120-150 kV for chest and 70-80 kV
for pelvis examinations. Corresponding recommendations by CEC (1996) are 125 kV (chest PA)
and 75-90 kV (pelvis AP). This should not impose any real problem provided the diagnostic
standard dose is well below the diagnostic reference level.
Questions about the validity of using VGAS and the CEC image criteria as a measure of clinical
image quality and of using the quantum noise SNR as a measure of physical image quality
were raised recently by Tingberg et al (2004) and Håkansson et al (2004). Tingberg et al
(2004) found no correlation between the results of a VGA-study and an FFE-study (freeresponse forced error; ROC-type of study) aiming at detection of pathological lesions in a
lumbar spine examination. Different dose levels at the detector were simulated by changing
the image noise. Håkansson et al (2004) found that the threshold contrast for detecting a 10
mm large simulated lesion (projected area) in different regions in the chest image had no
correlation to the expected contrast and scatter-to-primary ratio in that region. In fact, in the
region with the highest expected quantum SNR of the relevant details (Ullman et al 2004)
(hilar regions) required the largest threshold contrast (Håkansson et al 2004) indicating that
the observers found it more difficult to detected the detail in the hilar region. Sandborg and
Önnerth (2004) studied the human observer detection efficiency in two regions of a pelvis
phantom. In one region the inserted detail was projected on a homogeneous anatomical
background (colon) and in the other region (lower lumbar spine) the projected anatomy was
heterogeneous. The corresponding human detection efficiencies were 13% and 0.5%,
respectively, indicating that the detail in the heterogeneous background was much more
difficult to detect. These studies imply that detection of lesions may not be limited by quantum
noise but by the anatomical background structures, i.e. the patient. The implications of this are
at the moment not known in detail, but will be investigated further by for example including
measures of anatomical noise (Tischenko et al 2003) and anatomical background (Burgess et
al 2001) in the model of the signal-to-noise ratio expresssion.
Conclusions
Results by Tingberg and Sjöström (2003) and results from this study indicate that, with
modern digital imaging systems, it would be favourable to use lower tube voltages than
traditionally used with screen-film radiography. Image quality as measured by VGA analysis
and using the CEC image criteria clearly showed that image quality increases (the VGA score
-9-
(VGAS)) with decreasing tube voltage while maintaining effective dose. In this work, it has
been demonstrated that using an anthropomorphic phantom and a model of the imaging
system, physical measures of image quality can be calculated that correlate with the clinical
VGAS image quality descriptor used by Tingberg and Sjöström. This is encouraging since with
a model of the imaging process, extensive studies for optimisation can easily be made and
predict the performance of equipment not yet existing.
However, the significance of a clinical image quality descriptor based on VGA analysis applied
to the CEC image criteria (based on structures in the normal anatomy) may be questioned in
the light of recent research on the detectability of pathological lesions in clinical chest
(Håkansson et al 2004) and lumbar spine images (Tingberg et al 2004). While VGA analysis
and the CEC image criteria will give an overall evaluation of the characteristics of a good
image, the detection of pathological lesions depend also on other features in the image such as
anatomical noise and anatomical background structures.
Acknowledgements
This report was completed under the 5th framework program contract number FIGM-CT-200000036.
References
R. Birch, B. Marshall and G. M. Ardran. Catalogue of spectral data for diagnostic X-rays. The
Hospital Physicists' Association, Scientific Report Series 30, 47 Belgrave Square (London,
1979).
A E Burgess, F L Jacobson and P F Judy. Human observer detection experiments with
mammograms and power-law noise. Med Phys, 28, 419-437, 2001.
CEC European Commission. European Guidelines on quality criteria for diagnostic radiographic
images Brussels EUR 16260, 1996.
H G Chotas, C F Floyd, J T Dobbins 3rd, CE Ravlin. Digital chest radiography with
photostimuable storage phosphors: signal-to-noise ratio as a function of kilovoltage with
matched exposure risk Radiology 186, 395-398, 1993.
J T Dobbins 3rd, J J Rice, C A Beam, C E Ravin. Threshold perception performance with
computed and screen-film radiography: Implications for chest radiography, Radiology, 183,
179-187, 1992.
D Hart, DG Jones, BF Wall. Estimation of effective dose in diagnostic radiology from entrance
surface dose and dose-area product measurements NRPB-R262 Chilton: National Radiological
Protection Board, 1994.
M Håkansson, M Båth, S Börjesson, S Khedache, G Ullman, L-G Månsson. Nodule detection in
digital chest radiography: effect of nodule location (Submitted to Rad Prot Dosim)
ICRU, Photon, electron, proton, and neutron interaction data for body tissues, Report No. 46
International Commission on Radiation Units and Measurements (Bethesda, USA, 1992).
J H Launders, A R Cowen, R F Bury, P Hawkridge. Towards image quality, beam energy and
effective dose optimisation in digital thoracic radiography. Eur. Radiol., 11, 870-875, 2001.
G McVey, M Sandborg, D R Dance, and G Alm Carlsson. A study and optimisation of lumbar
spine x-ray imaging systems Br J Radiol, 76, 177-188, 2003.
- 10 -
M Sandborg D R Dance and G Alm Carlsson. Calculation of contrast and signal-to-noise
degradation factors for digital detectors in chest and breast imaging (Internal report 2003)
M. Sandborg, D. R. Dance and G. Alm Carlsson. Calculation of contrast and signal-to-noise
degradation factors for digital detectors in chest and breast imaging, ISRN ULI-RAD-R-93-SE,
2003.
M Sandborg, A Tingberg, D R Dance, B Lanhede, A Almén, G McVey, P Sund, S Kheddache, J
Besjakov, S Mattsson, L G Månsson, and G Alm Carlsson. Demonstration of correlations
between clinical and physical image quality measures in chest and lumbar spine screen-film
radiography British Journal of Radiology 74, 520-528, 2001.
M Sandborg, D R Dance, G Alm Carlsson and J Persliden. Monte Carlo study of grid
performance in diagnostic radiology: task dependent optimisation for screen-film imaging. Brit
J Radiol. 67, 76-85, 1994.
Michael Sandborg and Magnus Önnerth. Comparison of human observer efficiency in pelvis
radiographs in two different anatomical regions. ISSN 1102-1799, ISRN ULI-RAD-R--101—SE,
2004.
D Sjöström. Optimisation of an image plate system with respect to tube voltage – An observer
performance study based on chest and pelvis images (MSc thesis 2002)
SSI, Swedish Radiation Protections Institute’s regulations. Regulations and General Advice on
Diagnostic Standard Doses and Reference Levels within Medical X-ray Diagnostics (SSI FS
2002:2)
P Sund, C Herrmann, A Tingberg, S Kheddache, LG Månsson, A Almen et al. Comparison of two
methods for evaluating image quality of chest radiographs. Proc. SPIE 3981 251-257, 2000.
O. Tischenko, C. Hoeschen, O. Effenberger, S. Reissberg, E. Buhr, W. Doehring. Measurement
of the noise components in the medical x-ray intensity pattern due to overlaying,
nonrecognizable structures. Proc SPIE. Medical Imaging, 5030-43, 2003.
A Tingberg, C Herrmann, J Besjakow, K Rodenacker A Almen P Sund et al. Evaluation of
lumbar spine images with added pathology. Proc. SPIE 3981 34-42, 2000.
A Tingberg, D Sjöström. Search for optimal tube voltage for image plate radiography Proc.
SPIE, 5034, 187-196, 2003.
A Tingberg, M Båth, M Håkansson, J Medin, M Sandborg, G Alm Carlsson, S Mattsson and L-G
Månsson. Comparison of two methods for evaluation of image quality of lumbar spine
radiographs Proc. SPIE, 5372, 2004.
A Tingberg, C Herrmann, B Lanhede, A Almén1, M Sandborg, G McVey, S Mattsson, W Panzer,
J Besjakov, L G Månsson, S Kheddache, G Alm Carlsson, D R Dance, Ulf Tylén and M Zankl.
The influence of the characteristic curve on the clinical image quality of medical radiographs Br
J Radiol., 77, 204-215, 2004.
G Ullman, M Sandborg, D R Dance, R Hunt, and G Alm Carlsson. Optimisation of chest PA
using model calculations (Report 99).
G Ullman, M Sandborg, D R Dance, R Hunt, and G Alm Carlsson. The influence of patient
thickness, tube voltage and image detector on patient dose and detail signal to noise ratio in
digital chest imaging (Submitted to Rad Prot Dosim).
I. G. Zubal, C. R. Harrell, E. O. Smith, Z. Rattner, G. Gindi and P. B. Hoffer. Computerized
three-dimensional segmented human anatomy, Med Phys. 299-302, 1994.
- 11 -