Академический Документы
Профессиональный Документы
Культура Документы
discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/5503975
CITATIONS
READS
38
102
8 AUTHORS, INCLUDING:
Kathrin Yen
Christian Jackowski
Universitt Heidelberg
Universitt Bern
SEE PROFILE
SEE PROFILE
Michael Thali
Peter Vock
University of Zurich
Universitt Bern
SEE PROFILE
SEE PROFILE
ORIGINAL ARTICLE
20
medicine as a supplement to autopsy. Further radiologicpathologic case studies are necessary to dene the role of
postmortem CT and MRI as a single examination modality.
Key Words: virtopsy, virtual autopsy, postmortem, blunt chest
trauma, CT, MRI
(J Thorac Imaging 2008;23:2027)
J Thorac Imaging
Subjects
Between July 2000 and 2005, 30 forensic cases with
chest trauma were examined at our Institute of Forensic
Medicine in collaboration with the Department of
Diagnostic Radiology at the local University Hospital
using postmortem CT and MRI before autopsy. Six cases
that showed putrefaction (n = 5) or the ones not
examined using standard MRI sequences (n = 3) were
excluded from the study. Thus, 24 cases with chest trauma
that were examined using CT, standard MRI sequences,
and autopsy were included in this retrospective study.
The mean age of the 22 adult cases was 50 (age
range 18 to 80 y) and the remaining 2 cases were children
of 2 and 3 years of age. The mean weight of the 22 adult
cases was 75 kg (range 43 to 100 kg) and the mean height
was 174 cm (range 153 to 193 cm). The 2-year-old child
weighted 9 kg and was 71 cm tall; the 3-year-old child
weighted 12 kg and was 96 cm tall. There were 15 males
and 9 females in the group. The manner of death is
presented in Table 1. In all cases, the individuals died at
the accident scene or within the rst 2 to 3 hours in
hospital.
Autopsy
All 24 cases underwent conventional autopsy with
opening of all cavities approximately 12 hours after
radiologic scanning. Autopsy was carried out by 1 or 2
board-certied forensic pathologists, who obtained a
rough knowledge of preliminary radiologic diagnoses
supplied by a forensic resident, sometimes together with a
board-certied radiologist during the scan procedure.
RESULTS
Radiologic Evaluation Versus Autopsy Results
The results of the autopsy and radiologic examinations are presented in Table 1. In most of the cases, the
ndings were diagnosed with autopsy and using radiologic methods (Table 1). Some ndings, such as fractures
of the spine, mediastinal shift and the ndings of gas were
more frequently found via imaging rather than with
autopsy. In contrast, two-thirds of all cases with a heart
lesion remained radiologically undetected.
The number of the ndings diagnosed with autopsy
and by the radiologists is shown in Diagram 1. Except for
the ndings of diaphragm, the heart and the hemomediastinum, the number of ndings detected by imaging
methods was equal to or higher than the number
diagnosed with autopsy (Diagram 1).
The sensitivity and specicity of the results of the
rst and second radiologist in comparison with autopsy
are shown in Diagram 2. The sensitivity and specicity
regarding pneumomediastinum, pneumopericardium, and
21
Aghayev et al
22
TABLE 1. The Findings Diagnosed at Autopsy (A), by the First (1) and the Second Radiologist (2)
Manner of Death
Sex
Age
FatH
MusH
RFx
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
VA
VA
VA
Fall
BT
VA
VA
VA
VA
VA
Fall
VA
Fall
VA
VA
VA
VA
BT
VA
BT
VA
VA
VA
VA
M
M
M
M
M
M
M
M
F
M
M
F
F
F
M
F
M
M
M
F
F
F
M
F
47
67
25
33
67
40
3
52
56
32
53
65
65
40
17
2
54
80
61
42
75
52
34
47
A12
A12
2
A12
A12
A12
A12
A12
A12
A12
A12
A12
2
A12
A12
A12
A12
A12
A12
A12
2
2
A12
A12
12
2
A12
A12
A12
A12
A12
A12
A12
A12
2
2
A12
A12
12
A12
A1
A12
A12
2
A12
2
A12
A12
A12
A12
A12
A12
A12
A12
A12
A12
2
A12
A12
A12
A12
A12
A12
A12
A12
SFx
MedSh
PleEf
STEm
Pneu
PnMed
PnPer
PuLac
PuCon
PuAs
PuAt
PeEf
Heart
HeMed
12
A2
1
Diaph
A12
12
12
12
12
1
A
A12
A12
12
A12
A12
A12
A12
A12
12
A12
A12
A12
1
12
A12
A12
1
12
A1
12
A12
A12
A12
A12
2
A12
A12
A12
A12
A12
A12
A12
A12
A12
A12
A12
23
A12
A12
2
2
2
A12
A12
A12
A12
A12
A12
12
A12
A12
A12
A12
1
1
2
2
12
12
2
12
12
2
12
A12
A12
12
12
12
A12
A12
12
12
A12
A12
A12
A12
12
A12
A12
A12
A12
12
2
A12
A12
A12
A12
12
A12
A12
A12
12
2
A12
A
2
A12
2
A12
A12
A12
A12
1
A12
12
12
12
12
A12
12
12
A12
1
A12
1
A12
A12
12
A12
A12
A12
12
12
A12
A12
12
1
1
1
1
1
A12
A12
A12
12
A12
A12
1
A1
2
1
A12
A12
A12
A2
A12
A12
2
A12
A12
A12
A12
A12
A12
A12
A12
A12
A12
1
Aorta
A12
2
A12
A
1
1
A12
A12
A12
A2
A12
A12
A12
A12
A12
A12
A12
2
A2
A2
A1
A12
2
J Thorac Imaging
The ndings detected both at autopsy and using CT+MRI by both radiologists are additionally marked. In 18 cases out of 24, blunt chest trauma was caused by a vehicle accident, in 3 by a fall from height and in the
other 3 by blows with a body part.
Aorta indicates aortic rupture; BT, blunt trauma; Diaph, diaphragmatic rupture; FatH, fat tissue hemorrhage; Heart, heart injuries; HeMed, hemomediastinum; MedSh, mediastinal shift; MusH, muscle tissue
hemorrhage; N, numbers; PeEf, pericardial eusion; PleEf, pleural eusion; Pneu, pneumothorax; PnMed, pneumomediastinum; PnPer, pneumopericard; PuAs, pulmonary aspiration; PuAt, pulmonary atelectasis;
PuCon, pulmonary contusion; PuLac, pulmonary laceration; RFx, Rib fracture; SFx, spine fractures; STEm, soft tissue emphysema; VA, vehicle accident.
J Thorac Imaging
DIAGRAM 1. The diagram presents the sum of all findings detected at autopsy by the first and the second radiologist. Although
findings such as pneumomediastinum, pneumopericardium, and pericardial effusion were not documented at autopsy they were
seen using radiologic methods. Except for heart injuries, hemomediastinum, spine fracture, and diaphragmatic rupture, the
number of findings diagnosed radiologically exceeded the number diagnosed at autopsy.
Interobserver Correlation
The correlation between the ndings of the radiologists is presented in Diagram 2. The averaged
concordance was 90%. The correlation of only 55%
was seen in diagnosing pulmonary atelectasis and no
correlation was seen in pericardial eusion. In the
remaining ndings, a correlation equal to 75% or higher
was observed.
Summarizing the ndings of both radiologists
together and correlating them with autopsy results, a
sensitivity of 93% and a specicity of 59% were attained.
This improved average sensitivity at imaging by 3% to
4% but reduced average specicity by 7% to 16%.
DISCUSSION
Currently, CT and MRI are used more and more
often for postmortem forensic examination as methods
supplementing conventional autopsy. Within the scope of
validation of cross-sectional imaging in forensic medicine,
the question of benets and limitations of postmortem
CT and MRI arises.
It is already 3 decades that cross-sectional imaging
has been used for the clinical chest assessments; especially
CT is presently the method of choice in assessing chest
trauma patients. Recently, postmortem CT ndings of
the nontraumatic lung have been reported.8 Our
collected material permitted the comparison of results
23
Aghayev et al
J Thorac Imaging
DIAGRAM 2. The diagram shows the sensitivity and specificity of the first and second radiologist in comparison with autopsy.
24
specicities of the 2 radiologists in this regard. Interobserver dierences in the postmortem assessment easily
arise, for example, when one radiologist reads a small
mediastinal blister as a pneumomediastinum and the
other does not.
In clinical medicine, CT also is the method of choice
for detecting dicult rib and spine fractures.1113 The
high sensitivity of both radiologists in detecting bony
injuries in this study supports this statement. Again, the
dierences in specicity between the readers were unimportant and supposedly due to a dierent interpretation
of tiny ndings, for example, the inclusion of a fractured
spondylophyte among spine fractures.
One of the important advantages of postmortem
cross-sectional imaging of the body before autopsy is
the documentation of the full body in situ. By doing so,
the ndings can be assessed before any changes in the
location of the organs and tissues occur. This is primarily
important for ndings such as mediastinal shift and
pneumothorax, but also for pleural eusion (Fig. 1). It is
likely that both the appearance and the amount of these
ndings will be changed during the section of the thoracic
cavity, even before they have been recognized and
quantied. This may explain some of the dierences
between autopsy and radiologic examinations.
In detecting pulmonary laceration, contusion, and
aspiration, both radiologists showed a relatively high
sensitivity (89% to 100%) but a low-to-moderate
specicity (33% to 71%) (Diagram 2). This means that
radiologic recognition of pathologic changes to the lungs
is relatively easily performed but that dierentiation is
currently unsatisfying. This can be due to the nature of
these ndings. In laceration, contusion, and even aspiration, blood may be present in the alveoli of the lung, and
r
J Thorac Imaging
FIGURE 1. A, Tension pneumothorax (P) in the left thoracic cavity in a victim of blunt trauma. Note the mediastinal shift to the
right; B and C, Check for pneumothorax in the right thoracic cavity at autopsy in the same case. The usual preparation of a
window in the intercostal muscles and penetration of the parietal pleura are shown (arrow). The right lung sinks down indicating
no pneumothorax. It is well imaginable that the appearance of the mediastinal shift after use of this technique was changed owing
to the adjustment of the intrathoracic and ambient pressure. Thus, this usually used technique for the assessment of
pneumothorax might have contributed for the discrepancy in the imaging and autopsy results.
Interobserver Difference
In our assessment of chest trauma, we observed a
90% concordance between the rst and the second
radiologist. Knowing that postmortem radiology is a
very young domain combining forensic pathology and
clinical imaging methods and that in this study 2 clinically
experienced radiologists were employed, we estimate the
FIGURE 2. Pulmonary laceration (thick arrow), internal livores (double arrow) meaning livores within an organ as reported by
Jackowski et al, and blood aspiration to the main bronchus (arrow Bro) of a victim of blunt trauma. A, Axial CT image; B, Axial
T2-weighted MRI image; and C, Axial cut of the left lung after fixation in paraformaldehyde. Note pleural effusion (arrow PleEf)
in both radiologic images and the foci of aspiration in the upper left lobe on the autopsy image (dotted arrow).
r
25
Aghayev et al
Autopsy
Although ndings such as pneumomediastinum,
pneumopericardium, and pericardial eusion were not
mentioned at all in the autopsy protocols, they were
detected using combined radiologic methods. We suppose
that close radiologic-autopsy casework within the scope
of the evaluation of CT and MRI as noninvasive
examination methods in forensic medicine will lead to a
quality improvement of both forensic pathologic and
radiologic examinations.
In the present study, the averaged sensitivity of the
radiologic methods by each radiologist was at least 14%
higher than the corresponding specicity. This means that
in the radiologic interpretation there are more false
positive than false negative results; thus, ndings are
sometimes detected but their exact pathologic character is
unknown. However, cross-sectional radiologic-autopsy
correlation studies are still a young approach and, as
mentioned above, we believe that they oer a great
potential for both disciplines.
Furthermore, concerning the number of false
positive ndings responsible for the low specicity of
radiologic methods, sampling distances between tissue
cuts at autopsy are much larger than the 1.25 mm between
CT or the 5 mm between MRI images; there is a
possibility that small-sized ndings might have been
overlooked at autopsy. The use of supplemental minimally invasive postmortem examination techniques, such
as angiography and biopsy, might compensate for the
missing specicity.1417
As both radiologists were clinical specialists with
approximately 6 months experience in postmortem
cross-sectional imaging, we assume that an improvement
of their performance with greater experience is certain.
The best way to further train a radiologist is with
detailed analyses of autopsy images and ndings after
the radiologic assessment. Postmortem cross-section
radiology is a very young domain in medicine, which
is precisely in between forensic pathology and
radiology. Further broad research in this domain
with close radiologic-pathologic collaboration is
necessary.
Our study has the following limitations. First, the
autopsy protocols were used as a gold standard; autopsy,
however, uses subjective and observer-dependent assessment of the ndings in many respects. Second, we used
CT and MRI, which of course are expensive in routine
forensic work. MRI is more expensive, more timeconsuming, and less available, although it provides clearly
better contrast of soft tissues. CT is much faster and
less expensive; it is currently the method of choice
for diagnosis of bone and lung pathology and for the
detection of gas within the body. Finally, despite the
assumption that autopsy as the gold standard for
26
J Thorac Imaging
J Thorac Imaging
11.
12.
13.
14.
27