The British Journal of Radiology, 82 (2009), 157–161

SHORT COMMUNICATION

Evaluation of image quality and radiation dose in adolescent

thoracic imaging: 64-slice is preferable to 16-slice multislice CT
O J ARTHURS, MB, BChir, MRCPCH, PhD, S J YATES, MSc, CSci, MIPEM, P A K SET, MB, BS, FRCR, D A GIBBONS,
DCR(R), PGC and A K DIXON, MD, FRCR, FRCP

Department of Radiology, Cambridge University Teaching Hospitals NHS, Foundation Trust, Hills Road, Cambridge CB2
0QQ, UK

ABSTRACT. There is a constant drive for radiology departments to acquire newer and
improved CT machines in order to facilitate faster procedures and a greater repertoire
of examinations. However, it is unclear whether the newer technology provides
significantly improved image quality, or carries radiation dose implications for patients
during everyday clinical practice. We assessed image quality and radiation dose in 15
children and young adults aged 9.3–19.5 years who underwent thoracic imaging on
both 16-slice (16CT) and 64-slice (64CT) CT machines. Images were assessed for image
quality on a visual analogue scale (1 5 unacceptable; 5 5 perfect) and preferred image
set. All datasets were diagnostically acceptable (scores of 3 or more). The scores for
64CT datasets were significantly better than for 16CT datasets (mean scores of 4.5 and
4.0, respectively; p,0.05). The mean dose–length product (DLP) given was significantly
higher during 16CT examinations at 152 mGy cm (effective dose, 2.1 mSv) than for Received 17 August 2007
64CT examinations at 136 mGy cm (1.9 mSv; p,0.05). On average, 64CT examination Revised 10 March 2008
DLPs were 16 mGy cm (or 9%) lower than the equivalent 16CT examination DLPs. In the Accepted 19 March 2008
context of childhood and adolescent thoracic CT imaging, and using the same software
DOI: 10.1259/bjr/52970138
from the same manufacturers, 64CT examinations provide better image quality and
give a lower effective dose than do 16CT examinations. If the choice were available, it ’ 2009 The British Institute of
would be pertinent to use 64CT for this patient group. Radiology

There is growing concern over the radiation dose
received from CT imaging, with particular emphasis
being placed on patient groups with an increased
radiation sensitivity, such as children [1]. However,
radiation dose cannot be considered in isolation from
image quality, and there is a well-recognized trade-off
between image quality and radiation dose for CT [2].
Developments in multislice CT over the past decade
now allow large numbers of thin high-quality images to
be acquired in a very short examination time. This can
lead to improved resolution in the longitudinal direction
and reduced motion artefacts, and opens up new
applications for CT imaging. However, to maintain an
acceptable level of noise in narrower slices requires an
increase in patient dose. Furthermore, multislice CT can
intrinsically result in higher doses, because the X-ray
beam penumbra is not used for imaging. This was a
particular problem for early multislice systems, which
utilized a narrow total beam width [3]. However,
modern multislice systems also incorporate dose-saving
options, most notably automatic exposure controls, but
also features such as adaptive filtering for noise reduc-
Address correspondence to: Adrian Dixon, Department of
Radiology, Box 219, Addenbrooke’s Hospital, Cambridge
University Teaching Hospitals NHS Foundation Trust, Hills Road,
Cambridge CB2 0QQ, UK. E-mail: akd15@radiol.cam.ac.uk
tion. Factors affecting dose and image quality in modern
CT have been discussed in more detail elsewhere [4, 5].
There is a constant drive for radiology departments to
acquire newer and improved CT machines in order to
allow faster procedures and a greater repertoire of
examinations. However, the everyday clinical practice
implications of using newer machines, in terms of image
quality and the radiation dose to patients, remain
unclear.
There is some evidence that 64 multislice CT (64CT)
systems improve image quality over 16 multislice CT
(16CT) systems, particularly in cardiac imaging [6, 7].
One study of paediatric abdominal and pelvic examina-
tions has also suggested that 64CT systems provide
images that are of similar quality to other CT systems but
at significantly (up to 43%) lower radiation dose levels
[8].
In our institution, we currently have both 16CT and
64CT systems that are used for both adult and paediatric
imaging. In this article, we assess whether there is any
discernible difference between the diagnostic perfor-
mance of these two machines, using similar imaging
parameters in the same patient. We assess images from
children and young adults, as there is evidence of much
greater radiation susceptibility from CT, in terms of
increased lifetime risk per unit dose, in those under the
age of 20 years [9, 10].

The British Journal of Radiology, February 2009 157

 O J Arthurs, S J Yates, P A K Set et al

Methods Anonymized images were reviewed simultaneously

A 64CT machine was installed at our institution in
November 2004, alongside a 16CT machine. Over 600
multislice CT examinations were performed at our
institution on children and young adolescents (under
20 years of age) over a 2-year period up to November
2006. This included 204 thoracic CT examinations.
However, by chance alone, 34 patients had received
one thoracic CT examination on the 16CT machine
and one study on the 64CT machine. In our institution,
there is currently no system in place for choosing
or expressing a preference for either system: they
are allocated according to scheduling factors. 15 patients
had pairs of images that were suitable for review.
Standard clinical demographic details and measure-
ments were stored on an anonymized database, and
included age at examination, gender and patient weight.
Ethical permission was not required for this study, in
accordance with local research governance committee
guidance. We excluded pairs of images that differed in
reference mAs or the body parts imaged, or where limbs
had been included in one study but not the other (which
would require a greater X-ray penetration) owing to a
patient’s clinical condition.
We compared retrospectively the image quality and
radiation dose of image datasets acquired on a Somatom
Sensation 16 machine and a Somatom Sensation 64
machine (Siemens AG, Forchheim, Germany). The
technical characteristics of these CT systems and the
settings used in this study are given in Table 1 [11–13].
Both machines use the CARE Dose4DTM automatic
exposure control (AEC) system. This AEC system adjusts
the X-ray tube current to compensate for different levels
of beam attenuation along the longitudinal direction
during rotation, together with an adjustment for patient
size. It is controlled using a ‘‘reference mAs’’, which is
the mAs that would be used for a standard-sized patient.
The system adjusts the actual mAs delivered to provide
an appropriate (but not necessarily constant) level of
noise in the resulting image [14].
by two radiologists experienced in paediatric CT
practice, blinded to background clinical information,
CT machine, radiation dose and imaging parameters. All
thin slice width images comprising one complete dataset
were reviewed. Images were scored according to image
quality on a visual analogue scale:
1 5 unacceptable, no diagnosis possible;
2 5 marked artefacts, no diagnosis possible;
3 5 acceptable, diagnosis possible;
4 5 very good, diagnosis possible;
5 5 perfect, diagnosis possible.
The radiologists were then required to choose a
‘‘preferred image’’; both sets of images could receive
the same score, but one set had to be chosen for
preference.
Standard dose information, including the displayed
body volume CT dose index (CTDIvol) and dose–length
product (DLP), was recorded for each image set. The
DLP was converted into an estimate of effective dose
(ED) using a conversion factor published by the
National Radiological Protection Board [15]. The accu-
racy of the dose information reported by the machines
was assessed as part of routine quality assurance.
Averaged over the study period, measured values of
CTDIvol indices for the slice widths used in this study
were within 1.5% of displayed values for both
machines.
Statistical analysis was performed using a Wilcoxon
signed-rank test for numerical but non-continuous data
(image quality and preference data), whereas a paired
Student’s t-test was used for continuous, normally
distributed data (radiation dose) using SPSS 15.0 for
Windows (SPSS Inc., Chicago, IL). The binomial sign test
was used for non-parametric data (preferred data). All
tests were two sided and assessed at the 5% level of
significance. The null hypothesis tested was that there
was no difference between the 64CT and 16CT
machines.

Table 1. Technical characteristics and imaging parameters used [11–13]

16CT 64CT

Manufacturer Siemens Siemens

Type under comparison Sensation Somatom 16 Sensation Somatom 64
Detectors Solid-state ultra-fast ceramic Solid-state ultra-fast ceramic
Detectors per row 672 (1322 channels) 672 (1322 channels)
X-ray tube type/make Straton tube Straton tube
Total filtration at central axis (mm Al equivalent) 6.3 mm 6.8 mm
Total effective length of detector array at isocentre 24 mm 28.8 mm
Detectors 16 6 1.5 mm 32 6 0.6 mm
Beam collimation 24 19.2
kV 120 120
Reference mAs 100 100
Pitch 1.0 1.0
Measured body CTDIw at reference mAs 7.3 mGy 6.9 mGy
Reconstruction matrix 512 512
Adaptive filtration for noise Yes (automatic) Yes (automatic)
Cone beam correction SureView and AMPR SureView and AMPR
cone-beam artefact reduction cone-beam artefact reduction
Automatic mA control software CARE dose4DTM CARE dose 4D
CTDIw, weighted average CT dose index; AMPR, adaptive multiplane reconstruction.

158 The British Journal of Radiology, February 2009

Short communication: Evaluation of image quality and radiation dose: 64- vs 16-slice MSCT

Results

Demographics and imaging parameters

15 children and adolescents received CT examinations
of the thorax on both 16CT and 64CT machines. Nine of the
patients were male and six female. All had oncological
diagnoses, the most common of which was acute
leukaemia (five patients), but this sample group also
included soft-tissue tumours, lymphoma and bone
tumours. By chance, eight of the patients had their initial
examination on the 16CT machine, and vice versa for the
other seven patients. Mean age at CT examination was
13.3 years (standard deviation (SD), 3.1 years; range, 9.3–
19.5 years). There was no statistical difference between the
mean age for 16CT (13.3¡3.1 years) and that for 64CT
(13.3¡2.9 years; p50.71). Mean weight at CT examination
was 45.4 kg (SD, 17.4 kg; range, 18.1–75.5 kg). There was
no statistical difference between mean weight for 16CT
(45.9¡17.3 kg) and for 64CT (44.9¡17.9 kg; p50.26).

Image quality and preferred image

All 30 CT image sets (15 pairs) scored 3 or more for
‘‘adequate’’ diagnostic image quality. The radiologists
scored the 64CT examinations higher than the 16CT
examinations, with average scores of 4.5 and 4.0,
respectively (p,0.05; SD, 0.52 and 0.76, respectively;
Figure 1). Of the 15 64CT image sets, the radiologists
scored 7 as ‘‘4’’ and 8 as ‘‘5’’ whereas, of the 15 16CT
image sets, the radiologists scored 4 as ‘‘3’’, 7 as ‘‘4’’ and
4 as ‘‘5’’. In eight cases, the 64CT image sets scored
higher, and in two cases the 16CT image sets scored
higher. In five cases, there was no perceived difference.
When forced to choose, the radiologists preferred the
64CT image sets in 11 out of 15 cases (73%), and the 16CT
image sets in 3 cases (p50.06). In one case, no preference
could be made. In the five cases where both image sets
were scored equally, the 64CT was preferred in three
cases, with the 16CT preferred in one case and no
preference made in one of these cases.
An example of differences in image quality is given in
Figure 2.
Radiation dose
The mean recorded DLP was 144 mGy cm across all
examinations, i.e. an ED of 2.0 mSv [6]. The mean DLP
given during 16CT examinations was 152 mGy cm (stan-
dard error (SE), 14 mGy cm) and ED 2.1 mSv; during 64CT
examinations, DLP was 136 mGy cm (SE, 12 mGy cm) and
ED 1.9 mSv. This was a statistically significant difference
(p,0.05). Of 15 pairs of images, 11 recorded significantly
higher DLPs during 16CT than 64CT examinations. On
average, 64CT examination dose values were 16 mGy cm
(SE, 5.6 mGy cm); ED was 0.2 mSv (or 9%) lower than the
equivalent 16CT examination value.
There was a significant correlation between the 16CT
and 64CT DLP used (Figure 3). The correlation coeffi-
cient was 0.86, with a linear regression of 16CT DLP 5
1.1 6 64CT DLP + 0.96 mGy cm.
There were no significant difference in irradiated
patient length (p50.68) in this study. There was no
Figure 1. 64 multislice CT examinations scored higher than
16 multislice CT examinations (p,0.05) when images were
scored according to image quality on a visual analogue scale
(1 5 unacceptable, 2 5 marked artefacts, 3 5 acceptable, 4 5
very good, 5 5 perfect).
significant correlation between patient weight and
radiation dose (correlation coefficient, 0.38) or image
score (correlation coefficient, 0.24).
Discussion
Image quality
A previous study of seven patients suggested better
spatial resolution but no improvement in image quality
between 64CT and 16CT systems with a 43% lower
radiation dose [8]. Our study of 15 patients showed a
perceived improvement in image quality, with a much
more modest reduction in ED (9%). It is possible that
Thomas et al [8] used parameters which gave a lower ED
whilst compromising the image quality. Our study shows
that it is possible to get as good or better images from 64CT
at the same or a lower radiation dose than 16CT.
A lower ED at 64CT than 16CT has also been
demonstrated previously as a result of the introduction
of AEC at 64CT [16]. As our study uses the same AEC
software on both machines, any influence of AEC on the
results will be reduced.
Radiation dose
The radiation doses used in this study fell well within
the adult chest national reference dose third-quartile

The British Journal of Radiology, February 2009 159


Figure 2. An example of differences in image quality. A 10-year-old girl with Ewing’s sarcoma and previous pulmonary
metastases underwent unenhanced 2 mm slice thickness spiral (a) 16 multislice CT (16CT) and (b) 64 multislice CT (64CT)
examinations within a 3-month period to assess treatment response. Image quality was scored as ‘‘3’’ for the 16CT image and
‘‘5’’ for the 64CT image. The dose–length product was similar (72 mGy cm and 74 mGy cm, respectively) for this particular
example.
DLP value of 580 mGy cm (ED, 8.1 mSv) [15]. The
equivalent DLP for a 10-year-old child is 370 mGy cm
(ED, 4.8 mSv). The average ED during this study was
2.0 mSv, equivalent to approximately an extra 11 months
of background radiation [17]. We showed a difference in
ED between 64CT and 16CT of ,9%. Although in
absolute terms this may seem small, any step taken
towards reducing the radiation dose received by children
and young adults may result in significant socio-
economic savings in the future.
Figure 3. There was a significant correlation between the 16
multislice CT (16CT) and 64 multislice CT (64CT) doses used:
linear regression 16CT DLP 5 1.1 6 64CT DLP + 0.96 mGy cm.
The line of unity is also provided for reference.
160
O J Arthurs, S J Yates, P A K Set et al
Confounding factors/limitations of this study
Our results may be viewed with caution. Firstly, the
data sample size was small, and only machines made by
one manufacturer were used. This resulted from the
chance nature of using two different machines in our
department over the study time period, and the fact that
we did not expressly attempt to control for this by
exposing children to unnecessary additional radiation
purely for the purpose of this study. We were therefore
constrained by the retrospective nature of our study. A
randomized trial is one alternative way of approaching
this study.
Secondly, we only considered thoracic CT images. We
were unable to compare abdominal or pelvic CT
datasets, as our machines currently use different acquisi-
tion parameters to acquire these images and therefore
use different predetermined reference radiation doses.
Neither could we assess head CT examinations, as very
few head CT examinations had been repeated on patients
using two different machines, and neurological CT is
usually performed on a 4-slice CT at our institution. As
different doses are required to image different body
parts, it is difficult to extrapolate accurately these data to
make inferences across whole-body CT imaging.
Thirdly, one of the main determinants of patient dose in
this population is patient weight, so one of the potential
explanations for the dose difference observed could have
been because the patients happened to be consistently
heavier during imaging with one machine rather than the
other. Our results show that there was no difference
between patient weight between the two image acquisi-
tions, and no significant correlation between weight and
radiation dose (correlation coefficient, 0.38) or image score
The British Journal of Radiology, February 2009
Short communication: Evaluation of image quality and radiation dose: 64- vs 16-slice MSCT
(correlation coefficient, 0.24). Using AEC, a correlation
between weight and dose would perhaps be expected.
However, differences in abdominal fat content would
influence patient weight but would not affect the radiation
exposure in thoracic imaging, and might explain the lack
of correlation observed.
Conclusions
In the context of childhood and adolescent thoracic CT
imaging using the same software from the same
manufacturer, 64CT examinations provide better image
quality and give a lower ED when compared with 16CT
examinations. It is debatable whether these results can be
generalized to other manufacturers, but a real-life
comparison such as the one performed here would be
appropriate. We would encourage those looking to
investigate radiation dose and image quality parameters
on their own CT system to identify those acquisition
parameters which may be most amenable to change.
Acknowledgments
We are grateful to Dr Toby Prevost, Centre for Applied
Medical Statistics, Department of Public Health,
Cambridge, for invaluable statistical advice.
References
1. Frush DP, Donnelly LF, Rosen NS. Computed tomography
and radiation risks: what pediatric health care providers
should know. Pediatrics 2003;112:951–7.
2. ImPACT, Type Testing of CT Scanners: Methods and
Methodology for Assessing Imaging Performance and
Dosimetry. MDA Evaluation Report, MDA/98/25.
London, UK: Medicines and Healthcare Regulatory pro-
ducts Agency (MHRA), March 1998. Available from:
http://www.impactscan.org/report/MDA9825.htm.
3. Yates SJ, Pike LC, Goldstone KE. Effect of multislice
scanners on patient dose from routine CT examinations in
East Anglia. Br J Radiol 2004;77:472–8.
4. Lewis MA, Edyvean S. Patient dose reduction in CT. Br J
Radiol 2005;78:880–3.
5. ImPACT, Radiation dose issues in multi-slice CT scanning.
ImPACT technology update no. 3. London, UK: Medicines
and Healthcare Regulatory products Agency (MHRA),
January 2005. Available from: http://www.impactscan.
org/download/msctdose.pdf.
The British Journal of Radiology, February 2009
6. Seifarth H, Ozgun M, Raupach R, Flohr T, Heindel W,
Fischbacj R, et al. 64- versus 16-slice CT angiography for
coronary artery stent assessment: in vitro experience. Invest
Radiol 2006;41:22–7.
7. Zhang ZH, Jin ZY, Kong LY, Wang YN, Song L, Wang Y, et
al. Comparison of coronary artery bypass graft imaging
between 64-slice and 16-slice spiral CT. Zhongguo Yi Xue
Ke Xue Yuan Xue Bao 2006;28:21–5.
8. Thomas K, Matsumoto J, Fletcher J, McCollough C,
Bruesewitz M, Gregor N. Evaluation of image quality and
dose in a pediatric radiology practice: impact of 64-channel
mdct (64ct) systems. Proceedings of the annual meeting of
the Radiological Society of North America; 2005 November
27–December 5; Chicago, IL. Radiological Society of North
America, 2005: SSE15-01.
9. Brenner DJ, Elliston CD, Hall EJ, Berdon WE. Estimated
risks of radiation-induced fatal cancer from pediatric CT.
Am J Radiol 2001;176:289–96.
10. Recommendations of the ICRP. In: Smith H et al, editors.
Annals of the ICRP. Didcot, UK: Pergamon Press
1990;21:1–3,
11. ImPACT, Siemens SOMATOM Sensation 16. CT Scanner
Technical Evaluation. MHRA 04037. Reading, UK: NHS
Purchasing and Supply Agency, March 2004. Available from:
http://www.impactscan.org/report/MHRA04037.htm.
12. ImPACT, Sixteen slice CT scanner comparison report 14.
MHRA Report 06012. Reading, UK: NHS Purchasing and
Supply Agency, February 2006. Available from: http://
www.impactscan.org/report/Report06012.htm.
13. ImPACT, 32 to 64 slice CT scanner comparison report 14.
MHRA Report 06013. Reading, UK: NHS Purchasing and
Supply Agency, February 2006. Available from: http://
www.impactscan.org/report/Report06013.htm.
14. ImPACT, CT scanner automatic exposure control systems.
MHRA Report 05016. Reading, UK: NHS Purchasing and
Supply Agency, February 2005. Available from: http://
www.impactscan.org/report/Report05016.htm.
15. Shrimpton PC, Hillier MC, Lewis MA, Dunn M. Doses from
computed tomography (CT) examinations in the UK – 2003
review. Didcot, UK: National Radiological Protection Board
(now London, UK: Health Protection Agency), 2005: NRPB-
W67.
16. Greess H, Hendrich E, Baum U, Anders K, Brand M, Bautz
W. Daily, routine pediatric CT: examinations with auto-
matic exposure control. Proceedings of the annual meeting
of the Radiological Society of North America; 2005
November 27–December 5; Chicago, IL. Radiological
Society of North America, 2005: SSE15-03.
17. Watson S J, Jones A L, Oatway W B, Hughes J S. Ionising
radiation exposure of the UK population: 2005 review.
London, UK: Health Protection Agency, 2005: HPA-RPD-
001.
161