Seediscussions,stats,andauthorprofilesforthispublicationat:http://www.researchgate.net/publication/5503975

PostmortemImagingofBluntChestTraumaUsingCTandMRI:ComparisonWithAutopsy

ARTICLEinJOURNALOFTHORACICIMAGING·MARCH2008

ImpactFactor:1.74·DOI:10.1097/RTI.0b013e31815c85d6·Source:PubMed

CITATIONS

38

READS

102

8AUTHORS,INCLUDING:

KathrinYen

UniversitätHeidelberg

89PUBLICATIONS1,838CITATIONS

SEEPROFILE

ChristianJackowski

UniversitätBern

96PUBLICATIONS2,040CITATIONS

SEEPROFILE

MichaelThali

UniversityofZurich

408PUBLICATIONS4,773CITATIONS

SEEPROFILE

PeterVock

UniversitätBern

128PUBLICATIONS3,257CITATIONS

SEEPROFILE

Availablefrom:MichaelThali

Retrievedon:01November2015

Postmortem Imaging of Blunt Chest Trauma UsingCT and MRI

Comparison With Autopsy

Emin Aghayev, MD,*w Andreas Christe, MD,*z Martin Sonnenschein, MD,zy Kathrin Yen, MD,*Christian Jackowski, MD,* Michael J. Thali, MD,* Richard Dirnhofer, MD,*

and Peter Vock, MDz

Objective: Postmortem examination of chest trauma is an

important domain in forensic medicine, which is today

performed using autopsy. Since the implementation of cross-

sectional imaging methods in forensic medicine such as

computed tomography (CT) and magnetic resonance imaging

(MRI), a number of advantages in comparison with autopsy

have been described. Within the scope of validation of cross-

sectional radiology in forensic medicine, the comparison of

findings of postmortem imaging and autopsy in chest trauma

was performed.

Methods: This retrospective study includes 24 cases with chest

trauma that underwent postmortem CT, MRI, and autopsy.

Two board-certified radiologists, blind to the autopsy findings,

evaluated the radiologic data independently. Each radiologist

interpreted postmortem CT and MRI data together for every

case. The comparison of the results of the radiologic assessment

with the autopsy and a calculation of interobserver discrepancy

was performed.

Results: Using combined CT and MRI, between 75% and 100%

of the investigated findings, except for hemomediastinum (70%),

diaphragmatic ruptures (50%; n=2) and heart injury (38%),

were discovered. Although the sensitivity and specificity regard-

ing pneumomediastinum, pneumopericardium, and pericardial

effusion were not calculated, as these findings were not

mentioned at the autopsy, these findings were clearly seen

radiologically. The averaged interobserver concordance was

90%.

Conclusion: The sensitivity and specificity of our results

demonstrate that postmortem CT and MRI are useful

diagnostic methods for assessing chest trauma in forensic

medicine as a supplement to autopsy. Further radiologic-

pathologic case studies are necessary to define 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;23:20–27)

Approximately one-third of the time required for anautopsy is dedicated to examining the chest and one

of the important chest examination issues with autopsy isthe diagnosis of traumatic injuries. The majority of bluntchest trauma occurs due to motor vehicle accidents (90%)together with falls and work-related accidents (7%).1 Inthe last several years, the mortality rate for chest traumawas approximately 16%.2

The current gold standard for forensic postmortemchest examination is forensic autopsy. This allows for thedirect gaining of information by inspection and palpa-tion. The main limitations of autopsy are that it issubjective and observer-dependent. In addition, it canhardly be the basis for a second opinion owing to the factthat the topographic relations are changed and the bodytissues and organs are cut and cannot be stored for alonger time except for small specimens.

Presently, the use of postmortem computed tomo-graphy (CT) and magnetic resonance imaging (MRI) isgrowing in forensic medicine from year to year.3–6 Up tonow, publications on postmortem cross-sectional imagingin the literature either simply describe the postmortemradiologic appearance of forensically relevant find-ings,4,7,8 report on the abilities of CT and MRI inforensic routine9 or, rarely, present small studies on thecomparison between radiologic methods and autopsyregarding some specific forensic issues.5 However, forvalidation purposes of cross-sectional radiology inforensic medicine a direct comparison of the radiologicmethods and autopsy, aimed at determining the advan-tages and limitations of both methods, is necessary.

The aim of the following study was to evaluate theusefulness and to define the benefits and limitations ofpostmortem CT and MRI of chest injuries in forensicCopyright r 2008 by Lippincott Williams & Wilkins

From the *Institute of Forensic Medicine, University of Bern; wInstitutefor Evaluative Research in Orthopedic Surgery, MEM ResearchCenter; zInstitute of Diagnostic Radiology, Inselspital; andyDepartment of Diagnostic Radiology, Sonnenhof Spital AG,Bern, Switzerland.

Conflict of Interest: None of the authors has any conflicts of interest forthis study.

Reprints: Emin Aghayev, MD, University of Bern, Institute ofEvaluative Research in Orthopedic Surgery, Stauffacherstrasse 78,CH-3014 Bern, Switzerland (e-mail: emin.aghayev@memcenter.unibe.ch).

ORIGINAL ARTICLE

20 J Thorac Imaging � Volume 23, Number 1, February 2008

cases and to compare these with conventional autopsy inits role as the current gold standard.

MATERIAL AND METHODSThis study was approved by the responsible justice

department and also by the ethics committee of theUniversity of Bern.

SubjectsBetween July 2000 and 2005, 30 forensic cases with

chest trauma were examined at our Institute of ForensicMedicine in collaboration with the Department ofDiagnostic Radiology at the local University Hospitalusing postmortem CT and MRI before autopsy. Six casesthat showed putrefaction (n=5) or the ones notexamined using standard MRI sequences (n=3) wereexcluded from the study. Thus, 24 cases with chest traumathat 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 (agerange 18 to 80 y) and the remaining 2 cases were childrenof 2 and 3 years of age. The mean weight of the 22 adultcases was 75 kg (range 43 to 100 kg) and the mean heightwas 174 cm (range 153 to 193 cm). The 2-year-old childweighted 9 kg and was 71 cm tall; the 3-year-old childweighted 12 kg and was 96 cm tall. There were 15 malesand 9 females in the group. The manner of death ispresented in Table 1. In all cases, the individuals died atthe accident scene or within the first 2 to 3 hours inhospital.

Postmortem Cross-sectional ImagingPostmortem cross-sectional imaging of the body

using CT and MRI was performed before autopsy afterthe virtopsy approach.3 Radiologic examinations werecarried out at an average of 1.3 days after death (range 0to 5 d).

Full body CT scanning was performed on a 4-rowor 8-row scanner (Lightspeed QX/I unit, GE, USA) usinga collimation of 4� 1.25 or 8� 1.25mm, respectively.During the thorax CT scanning, both arms werecompletely elevated to avoid imaging artifacts fromextremity bones. A CT scan of the full body tookbetween 5 and 20 minutes. On a dedicated workstation,continuous multiplanar reconstructions using 1.25-mmreconstruction intervals were obtained. For case exam-inations, sagittal and coronal reformations were alsoused.

MRI of the thorax was carried out after CT on a1.5-T scanner (Signa Echospeed Horizon, 5.8 unit, GE,USA). Axial and coronal planes of the T2-weighted fastspin echo sequence [TE/TR 98/4000ms; slice thickness5mm (range 3 to 7mm); gap 1mm] with (n=22) and/orwithout (n=9) fat saturation were most frequently used.Axial planes of the T1-weighted fast spin echo sequence(TE/TR 14/400ms; slice thickness 5mm; gap 1mm)(n=11) and of the STIR sequence (TE/TR/TI 14/3000/130 or 22/4120/150ms; slice thickness 5mm; gap 1mm)

(n=8) were also obtained. MRI examinations of thethorax required between 45 minutes and 1.5 hours.

AutopsyAll 24 cases underwent conventional autopsy with

opening of all cavities approximately 12 hours afterradiologic scanning. Autopsy was carried out by 1 or 2board-certified forensic pathologists, who obtained arough knowledge of preliminary radiologic diagnosessupplied by a forensic resident, sometimes together with aboard-certified radiologist during the scan procedure.

Radiologic Evaluation and Data AnalysisThe radiologic data were independently evaluated

by 2 board-certified radiologists, blind to the autopsyfindings. Each radiologist interpreted postmortem CTand MRI data together for every case. Both radiologistswere specialists in clinical radiology with approximately 6months experience in postmortem CT and MRI imaging.

The diagnoses, which were evaluated in the post-mortem radiologic data of the chest, were the following:subcutaneous fat tissue hemorrhage (FatH), muscle tissuehemorrhage (MusH), fracture of the ribs or the sternum(RFx), the dorsal spine (SFx), mediastinal shift (MedSh),rupture of the diaphragm (Diaph), pleural effusion(PleEf), soft tissue emphysema (STEm), pneumothorax(Pneu), pneumomediastinum (PnMed), pneumopericar-dium (PnPer), pulmonary laceration (PuLac), pulmonarycontusion (PuCon), pulmonary aspiration (PuAs), pul-monary atelectasis including compression atelectasis(PuAt), pericardial effusion (PeEf), contusion or ruptureof the heart (Heart), hemomediastinum (HeMed), andrupture of the aorta (Aorta) (Table 1).

Statistical analyses of the results of the radiologicimaging with calculations of sensitivity and specificity incomparison with autopsy results were performed. Theinterobserver discrepancy was also assessed.

RESULTS

Radiologic Evaluation Versus Autopsy ResultsThe results of the autopsy and radiologic examina-

tions are presented in Table 1. In most of the cases, thefindings were diagnosed with autopsy and using radi-ologic methods (Table 1). Some findings, such as fracturesof the spine, mediastinal shift and the findings of gas weremore frequently found via imaging rather than withautopsy. In contrast, two-thirds of all cases with a heartlesion remained radiologically undetected.

The number of the findings diagnosed with autopsyand by the radiologists is shown in Diagram 1. Except forthe findings of diaphragm, the heart and the hemome-diastinum, the number of findings detected by imagingmethods was equal to or higher than the numberdiagnosed with autopsy (Diagram 1).

The sensitivity and specificity of the results of thefirst and second radiologist in comparison with autopsyare shown in Diagram 2. The sensitivity and specificityregarding pneumomediastinum, pneumopericardium, and

J Thorac Imaging � Volume 23, Number 1, February 2008 Postmortem Imaging of Blunt Chest Trauma

r 2008 Lippincott Williams & Wilkins 21

TABLE 1. The Findings Diagnosed at Autopsy (A), by the First (1) and the Second Radiologist (2)

N Manner of Death Sex Age FatH MusH RFx SFx MedSh Diaph PleEf STEm Pneu PnMed PnPer PuLac PuCon PuAs PuAt PeEf Heart HeMed Aorta

1 VA M 47 A12 A12 A12 A2 A12 12 12 12 12 12 12 1 A12 2 A2 VA M 67 2 A12 12 1 12 12 12 12 2 1 A12 12 1 2 A12 23 VA M 25 2 A12 12 2 A12 A14 Fall M 33 A12 A12 A12 12 1 A12 A12 A12 12 A12 A12 12 A125 BT M 67 A12 A12 A12 2 A12 A12 A126 VA M 40 A12 A12 A12 A12 2 2 A12 12 A12 2 A12 A12 A12 A12 A127 VA M 3 A12 A12 A1 A12 A12 A128 VA M 52 A12 A12 A12 12 A12 12 A12 1 A12 A12 A12 2 A A129 VA F 56 A12 A12 A12 12 A12 A12 A12 12 A12 A2 2 A1210 VA M 32 A12 2 A12 A12 A11 Fall M 53 2 2 A12 A12 12 1 12 A12 A12 1 A1212 VA F 65 2 2 2 A12 A12 2 A12 1 A213 Fall F 65 A12 A12 A12 A12 12 A12 A12 12 1 1 A12 A12 A12 A12 A12 A12 A1214 VA F 40 A12 A12 A12 A12 A12 12 12 A12 A12 A1215 VA M 17 12 12 A12 2 A12 12 A12 1 A12 A12 A12 1 A16 VA F 2 A1217 VA M 54 2 A12 A12 A12 A12 A12 12 12 1 12 A12 A12 2 1 A12 A1218 BT M 80 A12 A12 A12 A12 12 A12 A12 A12 A1219 VA M 61 A12 A12 A12 12 2 A12 A12 1 A12 A12 A12 A12 A1220 BT F 42 A12 A12 A12 A12 1 A12 A12 A1221 VA F 75 A12 A12 A12 A12 A12 A12 2 A12 12 1 A12 23 1 222 VA F 52 A12 A12 A12 A12 1 12 1 A1 A12 A12 A123 VA M 34 A12 A12 A12 12 A12 A12 A12 2 A12 A12 A12 A A2 A1224 VA F 47 A12 2 A12 A12 2 A A12 A12 A12 1 1 2 2 A2 2

The findings 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 theother 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 tissuehemorrhage; N, numbers; PeEf, pericardial effusion; PleEf, pleural effusion; 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.

Aghayev

etal

JThora

cIm

agin

g�

Volu

me

23,

Num

ber

1,

Feb

ruary

2008

22

r2008Lippincott

Willia

ms&

Wilk

ins

pericardial effusion were not calculated as these findingswere not found in autopsy protocols. The averagedsensitivity of the first radiologist was 89% and thatof the second was 90%. The averaged specificityof the first radiologist was 75% and that of the secondwas 66%.

The first radiologist detected all findings withsensitivity equal to or higher than 75% except forhemomediastinum (70%), diaphragmatic rupture (50%,n=2), and heart trauma (38%); the second radiologistdescribed all findings with sensitivity equal to or higherthan 75% except for diaphragmatic rupture (50%, n=2)and heart trauma (38%).

Regarding heart trauma, rupture of the heart withheart dislocation in 2 cases and a contusion in 1 case weredetected. On the other hand, rupture of the heartremained undiscovered in 5 cases and contusion in 1.

The specificity of the first radiologist for soft tissueemphysema, pulmonary aspiration, and pneumothoraxwas between 33% and 50% followed by pulmonarylaceration (57%), pulmonary contusion (60%), pulmon-ary atelectasis (72%), and spine fractures (73%). For theremaining findings, the specificity values were at least75% (Diagram 2).

The specificity of the second radiologist wasbetween 29% and 44% for muscle hemorrhage, pulmon-ary aspiration, and soft tissue emphysema. Pleuraleffusion and pneumothorax (50%), fat tissue hemorrhage(57%), rib fractures and pulmonary contusions (60%),mediastinal shift (65%), pulmonary lacerations (71%),and spine fractures (73%) followed. The specificity of thesecond radiologist for the remaining findings was equal toor above 75% (Diagram 2).

In summary, our results show that between 75%and 100% of autopsy findings can be discovered usingcombined CT and MRI examinations, except for hemo-mediastinum (70%), diaphragmatic rupture (50%;n=2), and heart injury (38%).

Interobserver CorrelationThe correlation between the findings of the radi-

ologists is presented in Diagram 2. The averagedconcordance was 90%. The correlation of only 55%was seen in diagnosing pulmonary atelectasis and nocorrelation was seen in pericardial effusion. In theremaining findings, a correlation equal to 75% or higherwas observed.

Summarizing the findings of both radiologiststogether and correlating them with autopsy results, asensitivity of 93% and a specificity of 59% were attained.This improved average sensitivity at imaging by 3% to4% but reduced average specificity by 7% to 16%.

DISCUSSIONCurrently, CT and MRI are used more and more

often for postmortem forensic examination as methodssupplementing conventional autopsy. Within the scope ofvalidation of cross-sectional imaging in forensic medicine,the question of benefits and limitations of postmortemCT and MRI arises.

It is already 3 decades that cross-sectional imaginghas been used for the clinical chest assessments; especiallyCT is presently the method of choice in assessing chesttrauma patients. Recently, postmortem CT findings ofthe nontraumatic lung have been reported.8 Ourcollected material permitted the comparison of results

DIAGRAM 1. The diagram presents the sum of all findings detected at autopsy by the first and the second radiologist. Althoughfindings such as pneumomediastinum, pneumopericardium, and pericardial effusion were not documented at autopsy they wereseen using radiologic methods. Except for heart injuries, hemomediastinum, spine fracture, and diaphragmatic rupture, thenumber of findings diagnosed radiologically exceeded the number diagnosed at autopsy.

J Thorac Imaging � Volume 23, Number 1, February 2008 Postmortem Imaging of Blunt Chest Trauma

r 2008 Lippincott Williams & Wilkins 23

of postmortem radiologic and autopsy examinations ofblunt chest trauma.

The pathologists performing the autopsies usuallyobtained a rough knowledge about the injured body partsas well as about some of the findings. One can supposethat the pathologists have had the benefit of the imagingdata. However, currently, autopsy is accepted as the goldstandard for postmortem chest examinations meaningthat autopsies detect the maximum findings. The aim ofthe present study was to evaluate the usefulness andto define the benefits and limitations of postmortem CTand MRI of blunt chest trauma in forensic cases incomparison with those of conventional autopsy, and theimaging methods were compared with the autopsyprocedures.

Radiologic Evaluation Versus AutopsyAlthough the formal specificity of both radiologists

was low for soft tissue emphysema, pneumothorax,and, partly, pneumomediastinum, we believe that thisis a problem of the verification standard. Autopsy, asgenerally known in clinical medicine and recentlyreported in the forensic context, misses many of thesefindings related to the presence of gas in the body thatcross-sectional radiologic methods and especially CT aresuperior in detecting, even in small amounts (Fig. 1).5,10

For the autopsy detection of such gas, for example,detection of gas within the heart, special techniques thatare not regularly performed are required, such as openingof the heart under water.11 We therefore assume that theautopsy findings with gas have been overlooked5,11 andthat this, by definition, artificially caused a low specificityfor these findings (Fig. 1). The fact that many of thesefindings are small may also explain the different

specificities of the 2 radiologists in this regard. Inter-observer differences in the postmortem assessment easilyarise, for example, when one radiologist reads a smallmediastinal blister as a pneumomediastinum and theother does not.

In clinical medicine, CT also is the method of choicefor detecting difficult rib and spine fractures.11–13 Thehigh sensitivity of both radiologists in detecting bonyinjuries in this study supports this statement. Again, thedifferences in specificity between the readers were unim-portant and supposedly due to a different interpretationof tiny findings, for example, the inclusion of a fracturedspondylophyte among spine fractures.

One of the important advantages of postmortemcross-sectional imaging of the body before autopsy isthe documentation of the full body in situ. By doing so,the findings can be assessed before any changes in thelocation of the organs and tissues occur. This is primarilyimportant for findings such as mediastinal shift andpneumothorax, but also for pleural effusion (Fig. 1). It islikely that both the appearance and the amount of thesefindings will be changed during the section of the thoraciccavity, even before they have been recognized andquantified. This may explain some of the differencesbetween autopsy and radiologic examinations.

In detecting pulmonary laceration, contusion, andaspiration, both radiologists showed a relatively highsensitivity (89% to 100%) but a low-to-moderatespecificity (33% to 71%) (Diagram 2). This means thatradiologic recognition of pathologic changes to the lungsis relatively easily performed but that differentiation iscurrently unsatisfying. This can be due to the nature ofthese findings. In laceration, contusion, and even aspira-tion, blood may be present in the alveoli of the lung, and

DIAGRAM 2. The diagram shows the sensitivity and specificity of the first and second radiologist in comparison with autopsy.

Aghayev et al J Thorac Imaging � Volume 23, Number 1, February 2008

24 r 2008 Lippincott Williams & Wilkins

additional differentiating features are rather discrete.Further autopsy-radiologic correlation studies on thedifferentiation of these findings are necessary (Fig. 2).

Both radiologists exhibited the same low sensitivityand high specificity for heart injuries. Two cases withventricle rupture and dislocation of the heart as well as 1case with ventricle contusion were detected. Two caseswith rupture of the heart ventricle, 2 cases with rupture ofheart atrium, and 1 case with heart contusion remainedradiologically undiscovered. Thus, our currently usedtechniques (CT with the collimation of 4� or8� 1.25mm, and 1.5-T MRI with the slice thickness of5mm) permit postmortem detection of heart injury inonly one third of the cases. This unsatisfying situationwas already mentioned in a previous study.3 Our resultsseem to indicate that rupture of the heart atrium is moredifficult to detect in postmortem CT or MRI incomparison to the rupture of the heart ventricle.Furthermore, both dislocations of the heart were seenby the radiologists, and we thus assume that its detectionis clearly possible using radiologic methods. In these 2

cases with heart dislocation, a rupture of the heartventricle was suspected and then confirmed using autopsyresults. These 2 detected ruptures of the heart ventriclewere the 2 largest ones, measuring between 3 and 4 cm atautopsy. Accordingly, the size of heart muscle ruptureplays a role in its detection using postmortem radiology.We assume that postmortem application of contrastmedia might significantly improve the detection of thetraumatic heart injury. According to a recent pilot studyin postmortem angiography using cross-sectional techni-ques, the meglumine-ioxithalamate as a contrast mediumpermitted an excellent visualization of the coronaryarteries at postmortem.14

Interobserver DifferenceIn our assessment of chest trauma, we observed a

90% concordance between the first and the secondradiologist. Knowing that postmortem radiology is avery young domain combining forensic pathology andclinical imaging methods and that in this study 2 clinicallyexperienced radiologists were employed, we estimate the

FIGURE 1. A, Tension pneumothorax (P) in the left thoracic cavity in a victim of blunt trauma. Note the mediastinal shift to theright; B and C, Check for pneumothorax in the right thoracic cavity at autopsy in the same case. The usual preparation of awindow in the intercostal muscles and penetration of the parietal pleura are shown (arrow). The right lung sinks down indicatingno pneumothorax. It is well imaginable that the appearance of the mediastinal shift after use of this technique was changed owingto the adjustment of the intrathoracic and ambient pressure. Thus, this usually used technique for the assessment ofpneumothorax might have contributed for the discrepancy in the imaging and autopsy results.

FIGURE 2. Pulmonary laceration (thick arrow), internal livores (double arrow) meaning livores within an organ as reported byJackowski et al, and blood aspiration to the main bronchus (arrow Bro) of a victim of blunt trauma. A, Axial CT image; B, AxialT2-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).

J Thorac Imaging � Volume 23, Number 1, February 2008 Postmortem Imaging of Blunt Chest Trauma

r 2008 Lippincott Williams & Wilkins 25

value of 90% as a fairly good interobserver concordance.It is probable that, with more experience in forensicpathology, both the accuracies of the individual readersand their concordance will improve.

AutopsyAlthough findings such as pneumomediastinum,

pneumopericardium, and pericardial effusion were notmentioned at all in the autopsy protocols, they weredetected using combined radiologic methods. We supposethat close radiologic-autopsy casework within the scopeof the evaluation of CT and MRI as noninvasiveexamination methods in forensic medicine will lead to aquality improvement of both forensic pathologic andradiologic examinations.

In the present study, the averaged sensitivity of theradiologic methods by each radiologist was at least 14%higher than the corresponding specificity. This means thatin the radiologic interpretation there are more falsepositive than false negative results; thus, findings aresometimes detected but their exact pathologic character isunknown. However, cross-sectional radiologic-autopsycorrelation studies are still a young approach and, asmentioned above, we believe that they offer a greatpotential for both disciplines.

Furthermore, concerning the number of falsepositive findings responsible for the low specificity ofradiologic methods, sampling distances between tissuecuts at autopsy are much larger than the 1.25mm betweenCT or the 5mm between MRI images; there is apossibility that small-sized findings might have beenoverlooked at autopsy. The use of supplemental mini-mally invasive postmortem examination techniques, suchas angiography and biopsy, might compensate for themissing specificity.14–17

As both radiologists were clinical specialists withapproximately 6 months experience in postmortemcross-sectional imaging, we assume that an improvementof their performance with greater experience is certain.The best way to further train a radiologist is withdetailed analyses of autopsy images and findings afterthe radiologic assessment. Postmortem cross-sectionradiology is a very young domain in medicine, whichis precisely in between forensic pathology andradiology. Further broad research in this domainwith close radiologic-pathologic collaboration isnecessary.

Our study has the following limitations. First, theautopsy protocols were used as a gold standard; autopsy,however, uses subjective and observer-dependent assess-ment of the findings in many respects. Second, we usedCT and MRI, which of course are expensive in routineforensic work. MRI is more expensive, more time-consuming, and less available, although it provides clearlybetter contrast of soft tissues. CT is much faster andless expensive; it is currently the method of choicefor diagnosis of bone and lung pathology and for thedetection of gas within the body. Finally, despite theassumption that autopsy as the gold standard for

postmortem chest examinations detect the maximumfindings, a potential bias in the study might result fromthe fact that the pathologists who carried out autopsyobtained a rough knowledge about the injured body partsand some of the radiologic findings.

The sensitivity and specificity of the results pre-sented here demonstrate that postmortem CT and MRIare useful diagnostic methods for assessing chest traumain forensic medicine. However, further radiologic-patho-logic case studies are necessary to define the exact role ofpostmortem CT and MRI as a single examinationmodality; currently, these methods should be used as asupplement to and not as a substitute of autopsy. Therecent and ongoing technical progress in radiologypromises a great potential for forensic cross-sectionalimaging.

ACKNOWLEDGMENTSThe authors are grateful to Elke Spielvogel, Carolina

Dobrowolska, and Christoph Laeser (Department ofRadiology, Bern University Hospital), Verena Beutlerand Karin Zwygart (Department of Clinical Research,Magnetic Spectroscopy and Methodology, Bern UniversityHospital) and also to Urs Koenigsdorfer and Roland Dorn(Institute of Forensic Medicine, Bern University) for theexcellent help in data acquisition during the radiologicexaminations and the forensic autopsies. Particular thanksgo to Eva Scheurer for distinguished support in thestatistical evaluation of the results.

REFERENCES1. Burckhard J, Engler C, Salinas L. La sante publique en Suisse,

prestations, couts, prix. Pharma Information. Basel: 1999. Availableat: http://www.interpharma.ch.

2. Wicky S, Wintermark M, Schnyder P, et al. Imaging of blunt chesttrauma. Eur Radiol. 2000;10:1524–1538.

3. Thali MJ, Yen K, Schweitzer W, et al. Virtopsy, a new imaginghorizon in forensic pathology: virtual autopsy by postmortemmultislice computed tomography (MSCT) and magnetic resonanceimaging (MRI)—a feasibility study. J Forensic Sci. 2003;48:386–403.

4. Jackowski C, Schweitzer W, Thali M, et al. Virtopsy: postmortemimaging of the human heart in situ using MSCT and MRI. ForensicSci Int. 2005;149:11–23.

5. Aghayev E, Yen K, Sonnenschein M, et al. Pneumomedia-stinum and soft tissue emphysema of the neck in postmortem CTand MRI; a new vital sign in hanging? Forensic Sci Int. 2005;153:181–188.

6. Aghayev E, Sonnenschein M, Jackowski C, et al. Postmortemradiology of fatal hemorrhage: measurements of cross-sectionalareas of major blood vessels and volumes of aorta and spleen onMDCT and volumes of heart chambers on MRI. AJR. 2006;187:209–215.

7. Shiotani S, Kohno M, Ohashi N, et al. Postmortem intravascularhigh-density fluid level (hypostasis): CT findings. J Comput AssistTomogr. 2002;26:892–893.

8. Shiotani S, Kohno M, Ohashi N, et al. Non-traumatic postmortemcomputed tomographic (PMCT) findings of the lung. Forensic SciInt. 2004;139:39–48.

9. Thali M, Vock P. Role of and techniques in forensic imaging. In:Payen-James J, Busuttil A, Smock W, eds. Forensic Medicine:Clinical and Pathological Aspects. London: Greenwich MedicalMedia; 2003:731–745.

10. Plattner T, Thali MJ, Yen K, et al. Virtopsy-postmortem multislicecomputed tomography (MSCT) and magnetic resonance imaging

Aghayev et al J Thorac Imaging � Volume 23, Number 1, February 2008

26 r 2008 Lippincott Williams & Wilkins

(MRI) in a fatal scuba diving incident. J Forensic Sci. 2003;48:1347–1355.

11. Jackowski C, Thali M, Sonnenschein M, et al. Visualization andquantification of air embolism structure by processing postmortemMSCT data. J Forensic Sci. 2004;49:1339–1342.

12. Aghayev E, Thali MJ, Sonnenschein M, et al. Fatal steameraccident; blunt force injuries and drowning in post-mortem MSCTand MRI. Forensic Sci Int. 2005;152:65–71.

13. Rivas LA, Fishman JE, Munera F, et al. Multislice CT in thoracictrauma. Radiol Clin North Am. 2003;41:599–616.

14. Jackowski C, Sonnenschein M, Thali MJ, et al. Virtopsy:postmortem minimally invasive angiography using cross section

techniques—implementation and preliminary results. J Forensic Sci.2005;50:1175–1186.

15. Aghayev E, Thali MJ, Sonnenschein M, et al. Post-mortem tissuesampling using computed tomography guidance. Forensic Sci Int.2007;166:199–203.

16. Aghayev E, Ebert L, Christe A, et al. CT data based navigation forpost-mortem biopsy—a feasibility study. Forensic Sci Int. 2007.Submitted for publication.

17. Jackowski C, Bolliger S, Aghayev E, et al. Reduction of post-mortem angiography induced tissue edema by using poly ethyleneglycol (PEG) as contrast agent dissolver. J Forensic Sci. 2006;51:1134–1137.

J Thorac Imaging � Volume 23, Number 1, February 2008 Postmortem Imaging of Blunt Chest Trauma

r 2008 Lippincott Williams & Wilkins 27