Forensic Science International 280 (2017) 8186

Contents lists available at ScienceDirect

Forensic Science International

journal homepage: www.elsevier.com/locate/forsciint

Comparative radiologic identication with standardized single CT

images of the paranasal sinusesEvaluation of inter-rater reliability
Cdric N. Bruna , Angi M. Christensenc , Milan Kravarskid,e , Guillaume Gorincourf ,
Wolf Schweitzera , Michael J. Thalia , Dominic Gaschoa , Gary M. Hatchg,
Thomas D. Rudera,b,*
a
Institute of Forensic Medicine, Department of Forensic Medicine and Imaging, University of Zurich, CH-8057 Zurich, Switzerland
b
Institute of Diagnostic, Interventional, and Pediatric Radiology, University Hospital Inselspital, University of Bern, CH-3010 Bern, Switzerland
c
Federal Bureau of Investigation Laboratory, Quantico, VA 22135, USA
d
Department of Orthopedic Surgery and Traumatology, Spital Tiefenau, CH-3004 Bern, Switzerland
e
Department of Orthopedic Surgery and Traumatology, Burgenspital Solothurn, CH-4500 Solothurn, Switzerland
f
LiiEEA 4264 (Experimental and Interventional Imaging Laboratory), CERIMED (European Center for Medical Imaging Research), Aix-Marseille University, F-
13005 Marseille, France
g
Center for Forensic Imaging, Departments of Radiology and Pathology, University of New Mexico School of Medicine, Albuquerque, NM 87102, USA

A R T I C L E I N F O A B S T R A C T

Article history:
Received 3 November 2016 The aim of this study was to assess the reproducibility of a standardized image for personal identication
Received in revised form 11 April 2017 (SIPI), used in the comparative analysis of paranasal sinuses, and test the effect of inaccurate reformation
Accepted 28 August 2017 of the SIPI on suitability for comparative identication.
Available online 2 September 2017 Five raters with different professional backgrounds independently reformatted SIPIs from ten post-
mortem head CTs. Inter-rater, intra-rater agreement as well angular deviations between reformatted SIPI
Keywords: images and reference SIPI images were calculated. Second, raters assessed the suitability of 70 accurately
Forensic radiology and inaccurately reformatted SIPIs for identication with a 4-point Likert scale. Inter-rater agreement as
Personal identication
well as levels of signicance regarding image suitability were calculated.
Radiologic identication
Inter-rater agreement regarding reproducibility of SIPI reformation was excellent (inter-rater
Paranasal sinuses
Standardized image for personal correlation coefcient (ICC) 0.9995, intra-rater ICC 0.9983). Deviation between the angular dimensions
identication of the reformatted SIPIs and the reference SIPIs was 1 in 94% of all 300 measurements. Inter-rater
agreement regarding the effect of inaccurate SIPI reformation on suitability for comparative
identication was fair (ICC 0.6809). There was no statistically signicant difference between raters
evaluation of image suitability (p = 0.9755).
This study shows that the standardized image for personal identication can be accurately reformatted
by different raters with varying professional backgrounds. In addition, raters agree that inaccurately
reformatted SIPIs are still suitable for comparative identication in the majority of cases.
2017 Published by Elsevier Ireland Ltd.

1. Introduction Technological developments in medical imaging and the subse-

There are several established and reliable methods to identify
the dead, including DNA samples, ngerprints or radiologic images
[1]. The process of identication depends on the scenario or
circumstances of death, the availability of experts [2], and the
availability and quality of antemortem records for comparison.
* Corresponding author at: Forensic Medicine and Imaging, Institute of Forensic
Medicine, University of Zurich, Winterthurerstrasse 190/52, CH-8057 Zurich,
Switzerland. Fax: +41 44 635 68 51.
E-mail addresses: thomas_ruder@hotmail.com, thomas.ruder@irm.uzh.ch
(T.D. Ruder).
quent ascent of cross-sectional imaging in forensic sciences
uncovered an enormous potential of radiologic images, especially
computed tomography (CT) for personal identication [24].
Irrespective of the medical imaging modality used for identica-
tion, an identication is performed through careful comparison of
ante-mortem (AM) images to post-mortem (PM) images, which
may include both macroscopic (visual) assessment as well as the
use of population frequency data for traits being compared [5].
Radiologic images of osseous structures are well suited for this
purpose because of their durability [6].
The rst published report on a positive identication with
radiologic images of the paranasal sinuses was published as early

http://dx.doi.org/10.1016/j.forsciint.2017.08.029
0379-0738/ 2017 Published by Elsevier Ireland Ltd.
82 C.N. Brun et al. / Forensic Science International 280 (2017) 8186

as 1927 [7]. The paranasal sinuses are fully developed by the age of
20 and their morphology remains unchanged throughout adult life
unless affected by pathologies [811]. In the late 1980s, Harris et al.
observed that the morphology of the frontal sinus features a high
inter-individual variability and hence offered a reliable means of
radiologic identication [12]. This conclusion was afrmed in
studies by Kullmann et al. and later by Christensen who used
Elliptical Fourier Analysis to demonstrate that the outline of the
frontal sinus is highly variable between individuals [1315]. A
signicant step towards modern radiologic identication was
taken in the early 1990s by forensic anthropologist Kathleen Reichs
who successfully identied a decedent based on her comparison of
ante-mortem CT (AMCT) and post-mortem CT (PMCT) of the
paranasal sinuses [16]. Since then, a number of publications have
demonstrated the utility and reliability of radiologic identication
based on CT images of the paranasal sinuses [2,6,17].
Today, there is general agreement that CT offers two funda-
mental advantages over conventional radiographs: First, CT images
are cross-sectional images and thus allow for a more detailed
comparison of the morphology of the paranasal sinuses than
conventional radiographs. Second, the use of multi-planar
reformation (MPR) tools enables investigators to reformat PMCT
images to match nearly any radiographic ante-mortem image [2].
Typically, alignment between PMCT and AMCT is achieved through
individual reformation of PMCT images to the axial AMCT images.
A recent publication proposed the use of a single standardized
image of the paranasal sinuses and orbit for identication [18]. The
use of standards in forensic sciences, particularly in comparative
identication can increase the accuracy of identications both in
individual cases as in mass disasters (quantitative identication
instead of qualitative visual comparison), decrease the time
needed to perform individual identications, and also improve
the evidentiary value of an identication in court (e.g. Daubert
Standard) [19,20].
In their article, Ruder et al. detail the method to create the
standardized image for personal identication (SIPI) from both
AMCT and PMCT data [18]. The SIPI is reconstructed parallel to the
orbito-meatal plane, using the inner ear and inferior margin of the
left frontal bone as anatomical landmarks [18]. A detailed
description of the anatomical landmarks is provided in Fig. 1.
They propose that this image plane is well suited for the purpose of
radiologic identication for the following reasons: (1) the selected
anatomical landmarks are covered in the scan volume of standard
head-CTs, which means that any existing clinical head-CT should
be suitable for comparison with a corresponding PMCT; (2) clinical
head-CT (i.e. potential AM reference material) is traditionally
acquired in the orbito-meatal plane, which means that the in-
plane resolution of an old non-isotropic CT image of the head will
be highest in the orbito-meatal plane and thus image quality
should be sufcient for comparative identication regardless of the
scan date of the AMCT; (3) all elements used for anatomical
orientation, reformation and identication are osseous, which
means that they should remain unchanged even after a long
postmortem interval; and (4) the selected anatomical landmarks
are thought to be unambiguous and easy to recognize, which
means that the reformation of the SIPI should be a straightforward
and simple task [18]. There is an abundance of published evidence
to support the authors rst three reasons for selecting the SIPI for
radiologic identication [2,17,2123]. However, the publication by
Ruder et al. does not provide any evidence to conrm the
supposedly straightforward and simple task of creating the SIPI.
The aim of this study is to (1) assess the reproducibility of the
SIPI by multiple raters with different professional backgrounds,
and (2) test the effect of inaccurate reformation of the SIPI on
suitability for comparative identication.
2. Methods
Ethical approval for this study was waived by the responsible
Ethics Committee of the Canton of Zurich, Switzerland (Waiver
Number: KEK Zurich 22-2015). PMCT images used in this study were
scanned between March 1 and March 4, 2016 as part of the judicial
investigation into the manner and cause of death of the deceased.
2.1. Study population
In this study, 10 postmortem head CT datasets were analyzed.
The population consisted of 7 males and 3 female with a mean age

Fig. 1. Axial and sagittal CT images of a skull. Fig. 1a corresponds to the standard image for personal identication (SIPI). The image plane of the SIPI (red dotted line in b) is
parallel to the orbito-meatal plane (blue dotted line in b) using the inner ear (yellow circles in a) and the inferior orbital margin of the left frontal bone (red arrows in b and a)
as anatomical landmarks. The center of cross-hairs on the axial image (Fig. 1a) is positioned over the small bone between the lateral semilunar canal and the vestibule The
anterio-posterior cross-hair line (red dotted line in b) is positioned just below the inferior left orbital margin of the frontal bone (red arrow in b). Note the numbers in the
lower left corner of the images (blue circles in a and b) which are used to document the angular position of the images within the x-, y-, and z-axis of the CT coordinate system.
(For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)
 C.N. Brun et al. / Forensic Science International 280 (2017) 8186 83

of 62.8 years (minimum 44 years; maximum 85 years; median 64.5
years). None of the cases featured traumatic or non-traumatic
pathologies of the sinuses or skull. Raters were blinded to all
patient information.
2.2. PMCT scan protocol
PMCT of the head was performed on a 128-slice Dual Source CT
scanner (Somatom Denition Flash, Siemens Healthcare, Erlangen,
Germany). Scan parameters of data used in this study were
(adapted from [24]):120 kVp; automatic dose modulation (CARE
Dose4D, Siemens Healthcare, Erlangen, Germany); 128  0.6 mm
collimation. PMCT image reconstruction was performed with a
slice thickness of 0.6 mm in increments of 0.4 mm in an adjusted
eld of view using soft tissue and bone-weighted tissue kernels in
all cases. For this study, only bone-weighted tissue kernel images
were evaluated. The viewing window was set to center of
1000 Hounseld Units (HU) and width of 4000 HU.
2.3. Image reformation and analysis
PMCT reformation and analysis was performed by ve raters
with different professional backgrounds (one radiologist, one
forensic pathologist, one forensic anthropologist, one orthopedic
surgeon, and one medical student) on a digital picture archiving
and communication system (PACS) viewer using a dedicated
visualization software (MM Reading, Syngo. Via VB10, Siemens
Healthcare, Erlagen, Germany). All raters underwent the same
training provided by the study supervisor (a board certied
radiologist) on how to operate the PACS viewer and the multi-
planar reformation (MPR) tool in three training cases. The study
supervisor did not participate in the study as a rater.
2.4. Reproducibility of the SIPI
To test the reproducibility of the SIPI, raters were asked to
separately and independently reformat each head CT with the
MPR tool and recreate an accurate SIPI according to the method
described by Ruder et al. in each of the 10 cases [18]. The
goniometry application of the MPR tool was used to document the
exact angular position of each reformatted SIPI with six angular
dimensions within the x-, y-, and z-axis of the CT coordinate
system (two angular dimensions for each axis) (Fig. 1). Consis-
tency of angular dimensions was assessed between all ve raters
as well as between all reformatted SIPIs and a set of 10 reference
SIPIs, previously reconstructed by the study supervisor. The time

Fig. 2. Sample case of inaccurately reformatted SIPI images. To assess the effect of inaccurate reformation of the SIPI on suitability for personal identication, the study
supervisor reformatted series of SIPI with lateral, cranial, and caudal rotation and three different degrees of angular deviation from the orbito-meatal plane.
84 C.N. Brun et al. / Forensic Science International 280 (2017) 8186
needed to reformat the SIPIs was documented for each rater. All
raters completed their individual image reformations in one
single session. In addition, the study supervisor who created the
reference SIPSs (and who is not included as one of the ve study
raters) repeated the reformation of all 10 SIPIs after an interval of
several months to assess the intra-rater agreement of the SIPI
reformation.
2.5. Effect of inaccurate reformation of the SIPI on suitability for
identication comparison
To assess the effect of inaccurate image reformation on the
suitability of the SIPI for comparative identication, raters were
independently asked to evaluate as series of 70 reformatted SIPIs
which included both accurately and inaccurately reformatted
images. Each stack of 70 images consisted of 7 accurately
reformatted SIPIs (taken from the reference SIPIs) as well as
63 inaccurately reformatted SIPIs with either a lateral, cranial, or
caudal rotation in predetermined angular degrees of 1, 3 , and 5 ,
respectively (Fig. 2). Raters used a 4-point Likert-scale to rate their
visual assessment of the effect of in-plane rotation and angular
deviation on image quality as compared to the reference SIPI
(adapted from [1]): (1) no deviation from SIPI (excellent suitability
for comparative identication); (2) minor deviation from SIPI
(good suitability for comparative identication); (3) moderate
deviation from SIPI (fair suitability for comparative identication);
(4) major deviation from SIPI (image not suitable for comparative
Fig. 3. Comparison of one case of reformatted SIPI images as reformatted by the study supervisor (reference SIPI) and all ve raters. The anatomical details of the paranasal
sinuses of all six images are perfectly aligned.
identication). The evaluations were examined for inter-rater
agreement as well as signicant levels of differences between
raters. The effect of in-plane rotation (i.e. lateral, cranial, and
caudal rotation) and the effect of angular deviation (in steps of 1,
3 , and 5 ) were separately assessed.
2.6. Statistical analysis
Continuous variables are expressed as means  standard
deviations or medians and ranges. Categorical variables are
expressed as frequencies (%). Inter-rater agreement regarding the
angular position of the reformatted SIPIs in the CT-coordinate
system (i.e. reproducibility of SIPI) as well as inter-rater
agreement regarding the effect of inaccurate reformation of the
SIPI on suitability for comparative identication was assessed
with the intra-class correlation coefcient (ICC). Intra-rater
agreement regarding the reproducibility of the reference SIPI
was also assessed with ICC. According to Landis and Koch, ICC
values of 1.0 indicate absolute, 0.80.99 excellent, 0.60.79 fair,
0.20.59 moderate, and 0.00.19 slight agreement [25]. The
KruskallWallis test was used to assess signicant levels of
differences between the raters evaluation of image suitability. A
p value <0.05 indicates statistical signicance. Consistency
between angular dimensions of the 10 reference SIPI and the
reformatted SIPI was calculated in differences of angular degrees.
Data analysis was performed using an online statistical software
tool (StatsToDo, Brisbane, Australia).
 C.N. Brun et al. / Forensic Science International 280 (2017) 8186
Fig. 4. Bar chart to visualize the high consistency regarding angular dimensions
between the reformatted SIPIs and the reference SIPIs. Angular dimensions
between the reformatted SIPIs and the 10 reference SIPIs were consistent (0
deviation) in 70% (211/300) of all angular measurements. Angular dimensions of the
reformatted SIPIs deviated from the reference SIPIs by 1 in 24% (71/300), by 2
in 4% (11/300), by 3 in 2% (6/300), and by 4 in 0.3% (1/300) (Fig. 4).
3. Results
Inter-rater agreement regarding the angular position of the
reformatted SIPI in the CT-coordinate system (reproducibility of
SIPI) was excellent (ICC 0.9995) (Fig. 3). Intra-rater agreement
regarding the reproducibility of the SIPI was also excellent (ICC
0.9983). Angular dimensions between the reformatted SIPIs and
the 10 reference SIPIs were fully consistent (0 deviation) in 70%
(211/300) of all angular measurements. Angular dimensions of the
reformatted SIPIs deviated from the reference SIPIs by 1 in 24%
(71/300), by 2 in 4% (11/300), by 3 in 2% (6/300), and by 4 in
0.3% (1/300) (Fig. 4). Mean time to reformat the SIPI was 3.3 min
per case (range 2.0 to 4.5 min, median 3.1 min). Inter-rater
agreement regarding the effect of inaccurate SIPI reformation on
suitability for comparative identication was fair (ICC 0.6809). The
KruskallWallis test revealed that there was no statistically
signicant difference between raters evaluation of image suitabil-
ity (p = 0.9755). Regarding in-plane rotation, SIPIs were rated
suitable for comparative identication (i.e. Likert scores of 13) in
92% (97/105) of all images with a lateral rotation, 59% (62/105) of
all images with a cranial rotation, and 53% (56/105) of all images
with a caudal rotation. Regarding angular deviation, SIPIs were
rated suitable for comparative identication (i.e. Likert score of
13) in 95% (100/105) of images with an angular deviation of 1 (in
any image plane), 62% (65/105) of images with an angular
deviation of 3 , and 48% (50/105) of images with an angular
deviation of 5 (Fig. 5).
4. Discussion
This study shows that the standardized image for personal
identication can be accurately reformatted by different raters
with a varying professional background. In addition, raters agree
that inaccurately reformatted SIPIs are still suitable for compara-
tive identication in the majority of cases.
85
Fig. 5. Bar chart to visualize the effect of inaccurate SIPI reformation on suitability
for comparative identication.
The fact that there was excellent inter- and intra-rater
agreement regarding SIPI reformation, regardless of the varying
professional background and experience with CT of all raters,
supports the claim that the landmarks selected by Ruder et al. for
the SIPI are easy to recognize. The fact that the deviation between
the angular dimensions of all reformatted SIPIs and the reference
SIPIs was 1 in 94% of all 300 measurements further indicates that
the SIPI is a highly reproducible image. Moreover, the time needed
to reformat a single SIPI ranged from only 2 to 4.5 min, which in
some of the authors experience is shorter than the time needed to
match a non-standardized MPR reformation of the head with
AMCT images. In other words, the SIPI can be accurately and
rapidly reformatted following the step-by-step instructions
published by Ruder et al. [18].
High reproducibility of accurately reformatted images and
short reformation time are also two important features of the SIPI
with regard to the work ow in disaster victim identication (DVI).
The high consistency of angular dimensions between the
reformatted SIPIs and the reference SIPI is relevant in view of the
assessment of image quality: raters agreed that image suitability of
incorrectly reformatted SIPIs with an angular deviation of 1 (in
any direction) was still adequate in 95% of all cases. In cases with an
angular deviation of 3 (in any direction), the SIPI was considered
suitable for comparative identication in 62%. Although this
nding clearly shows that angular deviations of 3 (and more) have
a considerable effect on the potential suitability of the SIPI for
comparative identication, one has to keep in mind that angular
deviations 3 occurred in only 6 of all 300 angular measurements
(2%). This means that the high reproducibility of the SIPI
compensates for the apparent susceptibility of image suitability
to angular deviations. In addition, a closer look at the data shows
that the effect of angular deviation on suitability of the SIPI is
different depending on the direction of in-plane rotation: there is
agreement among the raters that the effect of a lateral rotation on
image suitability is smaller than that of cranial or caudal rotation.
Therefore, inaccurate SIPI reformation is nearly always tolerated if
the deviation from a technically accurate reformatted SIPI either
occurs as lateral in-plane rotation or is 1 angular deviation in any
image plane.
This study has several limitations which deserve discussion.
The variable professional background may be seen as limitation
with regard to the second aim of the study (i.e. test the effect of
inaccurate reformation of the SIPI on image quality): the fact that
only two of the ve raters had practical experience with
comparative identication may have caused a greater heterogene-
ity regarding the assessment and rating of image quality. However,
it remains speculative if a more homogeneous group of raters with
respect to the extent of their practical experience in identication
would have rated the effect of inaccurate reformation on image
86 C.N. Brun et al. / Forensic Science International 280 (2017) 8186
quality as higher or as lower than the group of mixed raters
involved in this study. The authors hope to investigate this
question in a future study. The authors also recognize that a study
population of 10 cases is relatively small; however, in view of the
near-perfect inter-rater and intra-rater agreement it is question-
able if a larger number of cases would have affected the outcome.
5. Conclusions
This study shows that the standardized image for personal
identication can be reformatted by different raters with a varying
professional background with near perfect accuracy. In addition,
raters agree that inaccurately reformatted SIPIs are still suitable for
comparative identication in the majority of cases.
Conict of interest
The authors have no nancial conicts of interest to disclose.
Disclaimer
The views expressed are those of the authors and do not
necessarily reect the ofcial policy or position of the FBI. Names
of commercial manufacturers are provided for identication
purposes only, and inclusion does not imply endorsement of the
manufacturer or its products or services by the FBI.
Acknowledgments
The authors express their gratitude to to Emma Louise Kessler,
MD, for her generous donation to the Zurich Institute of Forensic
Medicine, University of Zurich, Switzerland.
References
[1] Interpol, Disaster Victim Identication Guide (DVIGuide) (version 2014),
(2016) Available from: https://www.interpol.int/INTERPOL-expertise/Foren-
sics/DVI. (Accessed 20 October 2016).
[2] G.M. Hatch, F. Dedouit, A.M. Christensen, M.J. Thali, T.D. Ruder, RADid: a
pictorial review of radiologic identication using post mortem CT, J. Forensic
Radiol. Imaging 2 (2014) 5259.
[3] M. Baglivo, S. Winklhofer, G.M. Hatch, G. Ampanozi, M.J. Thali, T.D. Ruder, The
rise of forensic and post mortem radiologyanalysis of the literature between
the year 2000 and 2011, J. Forensis Radiol. Imaging 1 (2013) 39.
[4] B. Morgan, A. Alminyah, A. Cala, C. ODonnell, D. Elliott, G. Gorincour, P.
Hofman, M. Iino, Y. Makino, A. Moskata, C. Robinson, G.N. Rutty, A. Sajantila, J.
Vallis, N. Woodford, K. Wo zniak, M. Viner, Use of post mortem computed
tomography in Disaster Victim Identicationpositional statement of the
members of the Disaster Victim Identication working group of the
International Society of Forensic Radiology and Imaging, J. Forensic Radiol.
Imaging 2 (2014) 114116.
[5] A.M. Christensen, G.M. Hatch, Quantication of radiologic identication:
development of a population frequency data repository, Proceedings of the
Annual Meeting of the American Academy of Forensic Sciences, Las Vegas, NV,
2016.
[6] F. Dedouit, F. Savall, F.Z. Mokrane, H. Rousseau, E. Crubzy, D. Roug, N. Telmon,
Virtual anthropology and forensic identication using multidetector CT, Br. J.
Radiol. 87 (2014) 20130468.
[7] W.L. Culbert, F.M. Law, Identication by comparison of roentgenograms: of
nasal accessory sinuses and mastoid processes, J Am Med Assoc 88 (1927)
16341636.
[8] R.E. Figueroa, J. Sullivan, Radiologic anatomy of the frontal sinus, in: S.E.
Kountakis, B.A. Senior, W. Draf (Eds.), The Frontal Sinus, Springer, Berlin, 2005,
pp. 720.
[9] M. Yoshino, S. Miyasaka, H. Sato, S. Seta, Classication system of frontal sinus
patterns by radiography: its application to identication of unknown skeletal
remains, Forensic Sci. Int. 34 (1987) 289299.
[10] R. Camerieri, L. Ferrante, D. Mirtella, F.U. Rollo, M. Cingolani, Frontal sinuses for
identication: quality of classications, possible error and potential correc-
tions, J. Forensic Sci. 50 (2005) 770773.
[11] B.G. Brogdon, Radiological identication of individual remains, in: M.J. Thali,
M. Viner, B.G. Brogdon (Eds.), Brogdons Forensic Radiology, CRC Press, Boca
Raton, 2010, pp. 153176.
[12] A.M. Harris, R.E. Wood, C.J. Nortj, C.J. Thomas, The frontal sinus: forensic
ngerprint? A pilot study, J. Forensic Odontostomatol. 5 (1987) 915.
[13] L. Kullman, B. Eklund, R. Grundin, Value of the frontal sinus in identication of
unknown persons, J. Forensic Odontostomatol. 8 (1990) 310.
[14] A.M. Christensen, Assessing the variation in individual frontal sinus outlines,
Am. J. Phys. Anthropol. 127 (2005) 291295.
[15] A.M. Christensen, Testing the reliability of frontal sinus outlines in personal
identication, J. Forensic Sci. 50 (2005) 1822.
[16] K.J. Reichs, R.B.J. Dorion, The use of computerized axial tomography (CAT)
scans in the comparison of frontal sinus congurations, Can. Soc. Forensic Sci. J.
25 (1992) 116.
[17] T.D. Ruder, M. Kraehenbuehl, W.F. Gotsmy, S. Mathier, L.C. Ebert, M.J. Thali, G.
M. Hatch, Radiologic identication of disaster victims: a simple and reliable
method using CT of the paranasal sinuses, Eur. J. Radiol. 81 (2012) e1328.
[18] T.D. Ruder, C. Brun, A.M. Christensen, M.J. Thali, D. Gascho, W. Schweitzer, G.H.
Hatch, Comparative radiologic identication with CT images of paranasal
sinusesdevelopment of a standardized approach, J. Forensis Radiol. Imaging
7 (2016) 19.
[19] A.M. Christensen, The impact of Daubert: implications for testimony and
research in forensic anthropology (and the use of frontal sinuses in personal
identication), J. Forensic Sci. 4 (2004) 427430.
[20] A.M. Christensen, G.M. Hatch, B.G. Brogdon, A current perspective on forensic
radiology, J. Forensic Radiol. Imaging 2 (2014) 111113.
[21] CT Teaching Manual: A Systematic Approach to CT Reading, in: M. Hofer (Ed.),
4th edition, Thieme, Stuttgart, 2010.
[22] Y.I. Kim, K.J. Ahn, Y.A. Chung, B.S. Kim, A new reference line for the brain CT: the
tuberculum sellae-occipital protuberance line is parallel to the anterior/
posterior commissure line, AJNR Am. J. Neuroradiol. 30 (2009) 17041708.
[23] L.J. Yeoman, L. Howarth, A. Britten, A. Cotterill, E.J. Adam, Gantry angulation in
brain CT: dosage implications, effect on posterior fossa artifacts, and current
international practice, Radiology 184 (1992) 113116.
[24] P.M. Flach, D. Gascho, W. Schweitzer, T.D. Ruder, N. Berger, S.G. Ross, M.J. Thali,
G. Ampanozi, Imaging in forensic radiology: an illustrated guide for
postmortem computed tomography technique and protocols, Forensic Sci.
Med. Pathol. 10 (2014) 583606.
[25] J.R. Landis, G.G. Koch, The measurement of observer agreement for categorical
data, Biometrics 33 (1977) 159174.