Professional Documents
Culture Documents
https://doi.org/10.1007/s00068-018-1021-9
ORIGINAL ARTICLE
Received: 1 May 2018 / Accepted: 25 September 2018 / Published online: 28 September 2018
© Springer-Verlag GmbH Germany, part of Springer Nature 2018
Abstract
Purpose To compare thick (5 mm) and thin slice images (1.5 mm) of lung, soft tissue, and bone window in thoracoabdominal
trauma computed tomography.
Materials and methods 167 Patients that underwent thoracoabdominal trauma CT between November 2014 and December
2015 were included in the study. CT data were reconstructed in a transverse direction with 5 mm and 1.5 mm slice images
of lung, soft tissue, and bone window. Two blinded raters (radiologists) evaluated the collected data by detecting predefined
injuries in different organ areas. Reconstruction and evaluation times as well as detected injuries were noted and compared.
Results Reconstruction and evaluation times were significantly higher with 1.5 mm thin-slice images, and the effect strength
according to Rosenthal displayed a strong effect of 0.61 (< 0.1 small effect, 0.3 middle effect, and > 0.5 strong effect). Average
evaluation time differences were 62.7 s (33.9 s–91.5 s) in bone window between 1.5 mm and 3 mm for rater 1 (p < 0.001) and
71.4 s (43.1 s–99.7 s) for rater 2 (p < 0.001). Average time differences between 1.5 mm and 5 mm were 68,7 s (43.9 s–93.5 s)
for rater 1 and 75.3 s (44.7 s–105.9 s) for rater 2 in lung window (p < 0.001) and 66.6 s (28.8 s–104.4 s) for rater 1 and 114 s
(74.4 s–153.6 s) for rater 2 in soft-tissue window (p < 0.001). There was no significant difference regarding soft-tissue and
lung injuries, except non-significant improvement in the detection of bone fractures.
Conclusion Thin-slice images do not bring any significant benefit in thoracoabdominal trauma CT of soft-tissue and lung
injuries, but they can be helpful for the diagnosis of bone fractures and incidental findings.
13
Vol.:(0123456789)
188 L. Guchlerner et al.
to initiate immediate action. Longer evaluation times may and lung window and 3 mm for bone window as in-house
be obstructive in this emergency situation. Thin slices often standard. In addition, thin-slice axial images with 1.5 mm
find usage in extended exploration of small cysts, tumor were deployed as experimental for all window settings.
classification, and for seeking metastases [16–19], whereas Table 1 gives an overview of reconstruction parameters.
thick slices are normally utilized for the swift detection All planes and slice thicknesses were calculated out of raw
of trauma and other pathologies [17]. The question arose image data from a scan in a mixed phase with a collimation
whether thin- or thick-sliced computed tomography gives of 0.6 mm. 5 mm slices were calculated standardly out of
better results in terms of clinical applicability in emergency raw image data; 1.5 mm slices were reconstructed addition-
situations and accuracy of the final diagnosis [14, 17, 18]. ally by medical-technical assistants.
In our study, we retrospectively evaluated if 1.5 mm axial The reconstruction time of all picture data was manually
images provide diagnostic advantage over 5 mm images in measured with a stopwatch and noted for each patient in
the detection of thoracoabdominal injury regarding differ- standard and experimental slice thickness for all window
ent window settings (lung, soft tissue, and bone). In addi- settings.
tion, incidental findings and small osseous incidents were
examined.
Objective image analysis
13
Comparison of thick- and thin-slice images in thoracoabdominal trauma CT: a retrospective… 189
Subjective image analysis the beginning of the evaluation process of each set of patient
picture data, in different organ areas and slice thicknesses,
The subjective analysis was performed by two independent the rater gave a starting signal to the timekeeper who started
raters (radiologists) with 5 and 3 years of experience, respec- the stopwatch. The watch was stopped as soon as the rater
tively. The raters evaluated each image stack from both, signaled his completion of the evaluation (Fig. 1).
standard and experimental slice thickness, in a randomized
order, and had no contact with each other during the assess- Statistical analysis
ment of picture data. Between the first and the second read-
ing of matching CT images in 5 mm and 1.5 mm both raters Statistical analysis was carried out with the aid of BiAS soft-
underwent a 30- to 60-day interval to minimize recognition ware (BiAS 11.02 for Windows, Epsilon Verlag, Frankfurt,
effects. The two raters were blinded against personal patient Germany). Interrater analysis of trauma incidents was car-
data and the diagnosis of the first CT scan before recon- ried out using Cohen’s weighted kappa to show agreement
struction, so that neither rater knew trauma consequences of between the two raters during examination (k > 0.81 excel-
probands before the study began. Blinding against the slice lent agreement, k = 0.61–0.8 good agreement, k = 0.41–0.6
thickness of the rated picture data was not possible due to a moderate agreement, k = 0.21–0.4 fair agreement, and
much higher image volume in the thinner layers. Evaluation k < 0.20 poor agreement). McNemar’s test was additionally
time was manually recorded and the comparison results of used for interrater event verification with Zimmermann’s
predetermined organ systems were analyzed. The rating pro- effect strength ω2 (ω2 = 0.01 low, ω2 = 0.1 moderate, and
tocol (Table 2) with predefined pathologies was developed, ω2 = 0.25 great effect strength) and Chinn’s f/Cohen’s d
so that both raters could assess the same organ areas for a effect strength (Chinn/Cohen: d = 0.2 low, d = 0.5 moder-
comparable evaluation outcome. ate, and d = 0.8 great effect).
The raters reviewed CT data on axial images in lung, soft Wilcoxon matched-pairs test with Hodges–Lehmann
tissue, and bone window, and identified organ traumata of Assessor calculated the average evaluation time difference
different body regions (Table 2). The dichotomous decision between 1.5 mm CT data and 5 mm, respectively, 3 mm (in
whether trauma consequences are present in named areas bone window) for the analysis of interrater time variations. p
was noted. They could adjust brightness settings to their value under 0.05 was considered as a statistically significant
preferred range. The radiological report was treated as the difference in all the tests.
gold standard regarding all the findings.
The assessment time of each rater was manually measured Objective image data quality (compared using SNR) was
and noted throughout the evaluation of lung, soft tissue, and significantly lower in 1.5 mm images compared to stand-
bone window in thick and thin slices of every proband. At ard 5 mm images (p < 0.001). The average SNR was 8.64
13
190 L. Guchlerner et al.
13
Comparison of thick- and thin-slice images in thoracoabdominal trauma CT: a retrospective… 191
13
192 L. Guchlerner et al.
13
Comparison of thick- and thin-slice images in thoracoabdominal trauma CT: a retrospective… 193
13
194 L. Guchlerner et al.
13
Comparison of thick- and thin-slice images in thoracoabdominal trauma CT: a retrospective… 195
20. Eichler K, Marzi I, Wyen H, et al. Multidetector computed 29. Jónsson H, Bring G, Rauschning W, et al. Hidden cervical spine
tomography (MDCT): simple CT protocol for trauma patient. injuries in traffic accident victims with skull fractures. J Spinal
Clin Imaging. 2015;39(1):110–5. https://doi.org/10.1016/j.clini Disord. 1991;4(3):251–63.
mag.2014.09.011. 30. Deutsche Gesellschaft für Unfallchirurgie AWMF Kurzversion der
21. Frellesen C, Stock W, Kerl JM, et al. Topogram-based automated S3—Leitlinie Polytrauma/Schwerverletzten—Behandlung. http://
selection of the tube potential and current in thoraco-abdominal www.awmf.org/upload s/tx_szleit linie n/012-019k_S3_Polytr auma
trauma CT—a comparison to fixed kV with mAs modulation _Schwerverletzten-Behandlung_2017-03.pdf. Accessed 15 Nov
alone. Eur Radiol. 2014;24(7):1725–34. https://doi.org/10.1007/ 2017.
s00330-014-3197-7. 31. Monnin P, Sfameni N, Gianoli A, et al. Optimal slice thickness
22. Sangster GP, González-Beicos A, Carbo AI, et al. Blunt trau- for object detection with longitudinal partial volume effects in
matic injuries of the lung parenchyma, pleura, thoracic wall, and computed tomography. J Appl Clin Med Phys. 2017;18(1):251–9.
intrathoracic airways: multidetector computer tomography imag- https://doi.org/10.1002/acm2.12005.
ing findings. Emerg Radiol. 2007;14(5):297–310. https://doi. 32. Bogner V, Mutschler W, Biberthaler P. Der Faktor “Zeit”. Seine
org/10.1007/s10140-007-0651-8. Bedeutung in der Pathophysiologie und Therapie des Polytraumas
23. Rubin GD. Data explosion: the challenge of multidetector-row (The “time” factor. Its impact in pathophysiology and therapy of
CT. Eur J Radiol. 2000;36(2):74–80. multiple trauma). Unfallchirurg. 2009;112(10):838–45. https://
24. Soo G, Lau KK, Yik T, et al. Optimal reconstructed section doi.org/10.1007/s00113-009-1606-1.
thickness for the detection of liver lesions with multidetector 33. Gur D. Imaging technology and practice assessments: diagnostic
CT. Clin Radiol. 2010;65(3):193–7. https://doi.org/10.1016/j. performance, clinical relevance, and generalizability in a chang-
crad.2009.10.009. ing environment. Radiology. 2004;233(2):309–12. https://doi.
25. Abdelmoumene A, Chevallier P, Chalaron M, et al. Detection of org/10.1148/radiol.2332040563.
liver metastases under 2 cm: comparison of different acquisition 34. Gur D, Bandos AI, Fuhrman CR, et al. The prevalence effect
protocols in four row multidetector-CT (MDCT). Eur Radiol. in a laboratory environment: changing the confidence ratings.
2005;15(9):1881–7. https://doi.org/10.1007/s00330-005-2741-x. Acad Radiol. 2007;14(1):49–53. https : //doi.org/10.1016/j.
26. Plurad DS, Nielsen JS, Hancock J, et al. Concomitant rib frac- acra.2006.10.003.
tures and minor liver or spleen injuries in blunt trauma: what 35. Berbaum KS, El-Khoury GY, Ohashi K, et al. Satisfaction of
is the potential for missed diaphragmatic injuries? Am Surg. search in multitrauma patients: severity of detected fractures.
2010;76(4):380–4. Acad Radiol. 2007;14(6):711–22. https: //doi.org/10.1016/j.
27. Claydon J, Maniatopoulos G, Robinson L, et al. Challenges acra.2007.02.016.
experienced during rehabilitation after traumatic multiple rib 36. Berbaum KS. Satisfaction of search in osteoradiology. AJR
fractures: a qualitative study. Disabil Rehabil. 2017. https://doi. Am J Roentgenol. 2001;177(1):252–3. https://doi.org/10.2214/
org/10.1080/09638288.2017.1358771. ajr.177.1.1770252c.
28. Dussa CU, Soni BM. A hidden injury. Emerg Med J.
2004;21(3):390–1.
13