Forensic Imaging 40 (2025) 200617

Contents lists available at ScienceDirect

Forensic Imaging
journal homepage: www.sciencedirect.com/journal/forensic-imaging

Unveiling the diagnostic accuracy of PMCT: Detection of pneumonia

considering postmortem changes and time intervals
Max G. Mentink a,b,* , Bartholomeus G.H. Latten c,d, Frans C.H. Bakers a, Casper Mihl a,e ,
Lisa M. Hillen c,f , Paul A.M. Hofman b
a
Maastricht UMC, Department of Radiology and Nuclear Medicine, P. Debyelaan 25, 6229 HX Maastricht, The Netherlands
b
Maastricht University, Care and Public Health Research Institute, Minderbroedersberg 4-6, 6211 LK Maastricht, The Netherlands
c
Maastricht UMC, Department of Pathology, P. Debyelaan 25, 6229 HX Maastricht, The Netherlands
d
Netherlands Forensic Institute, Laan van Ypenburg 6, 2497 GB The Hague, The Netherlands
e
Maastricht University, CARIM school for Cardiovascular Diseases, Universiteitssingel 40, 6229 ER Maastricht, The Netherlands
f
Maastricht UMC, Department of Pathology, GROW School for Oncology and Reproduction, P. Debyelaan 25, 6229 HX Maastricht, The Netherlands

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

Keywords: Purpose: Postmortem CT (PMCT) faces challenges in assessing lung parenchyma due to images being acquired in
Diagnostic accuracy study expiratory state, leading to varying severity of pulmonary edema redistribution with gravity-dependent atten­
Pneumonia uation ranging from ground glass to full opacification. This retrospective study assessed the effect of gravity-
Postmortem computed tomography
dependent attenuation and the postmortem time interval (PTI) on the diagnostic accuracy of PMCT for detect­
Clinical postmortem radiology
Computed tomography
ing acute pneumonia.
Radiology Materials and methods: Deceased patients who underwent PMCT and autopsy were included. Consensus evalu­
Autopsy ations by two radiologists and two pathologists re-examined images and histological samples of separate lung
lobes. Scores were assigned for radiological and histological findings, including the presence of acute pneumonia,
gravity-dependent attenuation severity, and pulmonary edema. PTI was calculated and correlated with gravity-
dependent attenuation severity. Crosstabs were created to calculate diagnostic parameters.
Results: Fifty-seven patients were included, with four excluded and 44 fully opacified lung lobes. 168 lung lobes
remained for analysis. The average PTI was 22 hours and 47 min. A weak correlation was observed between PTI
and gravity-dependent attenuation severity (τb = 0.125, p = 0.016). Acute pneumonia prevalence was 24,4 %,
with sensitivity and specificity of PMCT for all lung lobes at 31,71 % and 85,83 %, respectively. PMCT performed
better in subgroups with none or slight gravity-dependent attenuation and in patients scanned within 16 hours
after death.
Conclusion: Interpretation of lung parenchyma with PMCT is challenging. Statistical power was limited due to a
limited sample size. PMCT is more suited for excluding acute pneumonia than detecting its presence. Prolonging
PTI should be avoided, as increasing gravity-dependent attenuation severity over time limits PMCT sensitivity.

Introduction These challenges result from two inextricable characteristics of post­

mortem imaging. Primarily, the lungs are depicted in a passive expira­
Background tory state, leading to diminished alveolar aeration and heightened
attenuation of the lung parenchyma [9–14]. Secondly, postmortem
Postmortem computed tomography (PMCT) has been described as a changes occur soon after death due to the different pressure gradient
valuable adjunct to conventional autopsy, serving as an acceptable between the pulmonary vasculature and the alveolar space. This phe­
alternative when consent for autopsy is not provided by the next of kin nomenon involves a modification of capillary permeability, initially
[1–3]. However, difficulties with identifying pulmonary pathologies in characterized in 1950 as postmortem pulmonary edema [15]. The
postmortem imaging have been reported by multiple authors [1,3–8]. general thought is that ante-mortem pulmonary edema redistributes in a

* Corresponding author at: Biesterweg 56, 5615AJ, Eindhoven, The Netherlands.

E-mail addresses: max.mentink@mumc.nl (M.G. Mentink), bart.latten@mumc.nl (B.G.H. Latten), fch.bakers@mumc.nl (F.C.H. Bakers), casper.mihl@mumc.nl
(C. Mihl), lisa.hillen@mumc.nl (L.M. Hillen), p.hofman@maastrichtuniversity.nl (P.A.M. Hofman).

https://doi.org/10.1016/j.fri.2024.200617

Available online 20 November 2024

2666-2256/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
M.G. Mentink et al. Forensic Imaging 40 (2025) 200617

distinct pattern in the dependent lung lobes when death occurs. In fact, questions [8,25,26]. Data was anonymized and handled according to
the association between ground glass attenuation and histopathological national legislation. Patients were only excluded based on absence of
pulmonary edema has been reported in several studies [16,17]. On CT reference standard.
imaging, this postmortem pulmonary edema can be recognized by a
bilateral and symmetrical gradient of gravity-dependent attenuation Procedures
[13,17]. This characteristic gravity-dependent attenuation pattern may
encompass ground-glass attenuation and consolidation, and is reported A full-body PMCT was performed as the index test. The PMCT was
to occur in up to 86 % of all cadavers [5,18,19]. While postmortem carried out within the first workday after death. PMCT was performed
changes are regarded not as pathology but as a consequence of gravity, with a Siemens SOMATOM Definition Flash or Philips Brilliance 64
hypostasis, and the cessation of tension forces, they have the potential to scanner. The scanner-specific imaging protocols have been published in
obscure or exaggerate antemortem pathology [13,18,20–22]. For this previous work [25]. The acquired images were viewed in appropriate
reason, postmortem changes have a substantial impact on the radiolo­ window-level settings under normal working conditions on diagnostic
gist’s interpretation of pulmonary findings on PMCT images. When an workstations. Two radiologists (with 6- and 3-years’ experience in
acute pneumonia is suspected, certainty is needed as this can alter the forensic and/or clinical PMCT, respectively) re-assessed the PMCT scans
cause of death, or falsely indicate an unrecognized and untreated clin­ and provided the scoring in consensus after cases were anonymized. The
ically relevant co-morbidity. Several research groups have explored the postmortem time interval (PTI) was defined as the time interval between
application of intermittent or continuous positive airway pressure the recorded time of death and the PMCT scan.
ventilation on cadavers. It is a method of simulated breath holding, Histology samples from four lung lobes (encompassing both upper
referred to as ventilated PMCT [20,21,23,24]. By applying high pressure and lower lobes) acquired during clinical autopsy according to the
to the airways, the airways and alveoli are aerated and the edema is standardized autopsy protocol were re-assessed for this study and were
subsequently reduced. Consequently, the underlying assumption is that used as reference standard for analysis. Histology samples of the middle
pathology becomes more discernible due to the ability to push sur­ lobe were not always available or labeled in the same standardized
rounding edema back into the capillaries, whereas pus, aspiration ma­ manner as the other lobed, therefore the middle lobes were not
terial and nodules will remain present [20,21,23]. The literature concluded in this analysis (missing reference standard). Two patholo­
currently lacks comprehensive insights into the diagnostic efficacy of gists, one specialized in pulmonary pathology and the other in autopsy
clinical PMCT for detecting acute pneumonia. Exploring the diagnostic pathology, collaboratively provided the consensus scoring. Macroscopic
value of PMCT is essential to assess the need for supplementary tech­ re-evaluation of the lung lobes could not be performed. The pathologists
niques, like ventilated PMCT, to enhance accuracy. Additionally, it is were blinded to the original autopsy report as well as the PMCT results
also of interest to know whether the severity of gravity-dependent after re-assessment, but not to the original request form, containing
attenuation and the postmortem time interval affect this diagnostic limited clinical information to simulate daily practice. If multiple
value. We hypothesize that the severity of pulmonary edema is inversely microscopic samples of a single lobe were present, then all samples were
proportional to the diagnostic value of PMCT in the detection of acute re-assessed. The original pathology report was solely consulted for lung
pneumonia. weight at the time of the autopsy.

Purpose Scoring

This study aimed to investigate the diagnostic value of PMCT in The thorax was scored for the presence of pleural effusion, defined as
detecting acute pneumonia, using histopathology diagnosis as the an abnormal collection of fluid in the pleural space. The right upper,
reference standard. Additionally, the study explored the impact of right lower, left upper, and left lower lung lobes were subsequently
postmortem changes’ severity and the time interval between death and scored for the presence of pre-existing pulmonary disease (e.g. emphy­
imaging on the diagnostic value. sema, interstitial lung diseases), the severity of gravity-dependent
attenuation (none, slight, advanced, severe, or fully opacified), and
Methods the presence of acute pneumonia. Fig. 1 shows the different severities of
gravity-dependent attenuation from slight to severe (none and fully
Study design opacified not depicted). Fully opacified lobes were excluded from
analysis concerning the detection of pneumonia as these were consid­
This retrospective observational study, evaluating the diagnostic ered as inconclusive. A five-point Likert scale was used to record the
efficacy of PMCT as an index test for diagnosing acute pneumonia, was radiologists’ estimates of confidence in the diagnosis of acute pneu­
conducted in the routine clinical setting of an academic tertiary hospital. monia, according to the following descriptors: definitely not (1), prob­
Histology samples served as the reference standard, with both PMCT ably not (2), uncertain (3), probable (4), and definite (5). Radiologists
scans and histology samples re-examined and scored. The study protocol were informed of the PTI, recent laboratory results (if available; C-
received a waiver of approval from the local ethics committee, because reactive protein and leucocyte count), and last measured temperature
there were no living human participants included in this retrospective within the last 48 hours before death if requested because the inter­
study. This research received no specific grant from any funding agency pretation of airspace opacification is subject to clinical and laboratory
in the public, commercial, or not-for-profit sectors. parameters and to simulate daily practice. In the appropriate clinical
setting, acute pneumonia was suspected when a patchy or confluent
Participants and data homogeneous increase in parenchymal attenuation attributed to the
obscuration of underlying vessels was observed, according to Shiotani
In the period of September 2015 until July 2017 PMCT was imple­ et al.17 The estimates ‘probable’ and ‘definite’ were considered as pos­
mented as a postmortem examination additional to autopsy. Patients itive diagnosis of acute pneumonia. When a lobe was fully opacified, the
submitted to a full-body PMCT prior to autopsy were included from the entire lung lobe was considered inconclusive.
wards of the Department of Internal Medicine in a consecutive series. Histological formalin-fixed and paraffin-embedded (FFPE) tissue
Consent for the postmortem examination was obtained from the next of samples stained with Hematoxylin and Eosin (H.E.) obtained during the
kin by the treating physician on the wards of the Department of Internal autopsy of the four lung lobes were re-assessed by two pathologists. The
Medicine. The data of the patients included in this analysis have in part pathologists systematically classified the specimens based on the
been analyzed and published previously with different research absence or presence of various pathological entities, including acute

2
M.G. Mentink et al. Forensic Imaging 40 (2025) 200617

Diffuse alveolar damage, a severe form of acute lung injury, was char­
acterized by the formation of hyaline membranes lining the alveolar
surfaces, reflecting disruption and damage to the alveolar-capillary
barrier. Fibrosis was defined as excessive deposition of collagen and
other extracellular matrix components in the lung tissue, thereby lead­
ing to replacement of preexistent alveolar architecture by scar tissue.
Pulmonary congestion described the accumulation of blood within the
pulmonary vessels. Emphysema characterized the destruction and loss
of lung tissue, particularly the alveoli, yielding in enlarged air spaces.
Pulmonary edema involved abnormal accumulation of fluid within the
lung interstitium and alveoli. Interstitial inflammation referred to
inflammation within the lung interstitium, involving the connective
tissue between air spaces.

Statistical analysis

A box plot was constructed to visualize the median values and the
interquartile ranges of the PTI of the defined subgroups of gravity-
dependent attenuation severity. When multiple variables were tested
for correlations, appropriate Bonferroni corrections were performed.
Different parameters of diagnostic value were calculated for the entire
sample size and several defined subgroups according to the severity of
gravity-dependent attenuation and PTI, including sensitivity, specificity,
positive predictive value (PPV), negative predictive value (NPV), and
accuracy. Statistical analyses were performed using SPSS (IBM® SPSS®
Statistics for Macintosh, Version 24.0.0.0. Armonk, NY: IBM Corp.). A p-
value of <.05 was considered statistically significant.

Results

Participants

A total of 57 patients underwent PMCT followed by clinical autopsy.

The histopathological slides of four patients were not available. There­
fore, these patients were excluded from further analysis. Four lobes were
scored separately, consequently, 212 lobes were available for analysis of
53 patients. Forty-four lung lobes were omitted from the analysis for
acute pneumonia detection, as a definitive radiological diagnosis could
not be confidently established due to complete opacification of the lung
lobes. One hundred sixty-eight lobes remained for analysis of diagnostic
value parameters. A flowchart depicting the patients who were included
and excluded is presented in Fig. 2. Patient, clinical, and radiological
characteristics are described in Table 1.

Postmortem time interval

The mean PTI was 22 h and 47 min (± 17:18). Forty-three percent of

patients were scanned between 12 and 24 h after death. Gravity-
dependent attenuation Kendall’s tau-b (τb) correlation coefficient was
calculated to determine the relationship between the PTI and severity of
gravity-dependent attenuation. There was a very weak positive corre­
lation between the PTI and severity of gravity-dependent attenuation
(τb = 0.125), which was statistically significant, p = 0.016 (CI 0.036-
Fig. 1. Examples of gravity-dependent attenuation severity, A = Slight, B = 0.212). Independent samples Kruskal Wallis H test showed a significant
advanced, C = severe. difference (p = 0.031) between the median PTI of the subgroups with no
gravity-dependent attenuation changes and fully opacified lung lobes
pneumonia, organizing pneumonia, diffuse alveolar damage, fibrosis, (after appropriate adjustment with the Bonferroni correction for multi­
malignancy, congestion, emphysema, edema, interstitial inflammation, ple tests). This result is in concordance with a very weak τb correlation.
alveolar hemorrhage, siderophages, and thrombi. These classifications A boxplot illustrating the PTI for each severity of GDA is included in
were graded on a scale of ‘none’, ‘minimal’, ‘moderate’, ‘severe’, or ‘not Fig. 3. Additionally, the proportion of patients with pleural effusion did
to be assessed’. Acute pneumonia was characterized by an inflammatory not differ between the two groups PTI ≤ 16 h and PTI > 16 h (44 % vs.
infiltrate containing neutrophilic granulocytes, whilst organizing 32 %, p = 0.374, respectively). Lung weight was not statistically
pneumonia exhibited fibro-histiocytic proliferations within the alveolar different between the two groups, for both the left side (730.68 gr vs.
spaces and interstitium, indicating a reparative response to the initial 752.44 gr, p = 0.483, respectively) and right side (898.88 gr vs. 898.44
lung injury. Both stages of pneumonia could coexist simultaneously. gr, p = 0.919, respectively).

3
M.G. Mentink et al. Forensic Imaging 40 (2025) 200617

(0.185), diffuse alveolar damage (0.255), fibrosis (0.187), malignancy

(0.004), congestion (0.202), emphysema (-0.19), edema (0.197), inter­
stitial inflammation (-0.131), alveolar hemorrhage (0.070), side­
rophages (-0.185) or thrombi (-0.037). None of the correlations
remained statistically significant after appropriate Bonferroni
corrections.

Detection of acute pneumonia

The overall prevalence of acute pneumonia was 24.4 %. Table 2

includes the sensitivity, specificity, positive predictive value, negative
predictive value, and diagnostic accuracy with corresponding confi­
dence intervals for the detection of acute pneumonia by PMCT. Several
subgroup analyses were performed such as subgroups by severity of
gravity-dependent attenuation (none combined with slight, advanced,
and severe), as well as patients with a PTI ≤ 16 h and PTI > 16 h. As
shown, the overall sensitivity of PMCT for the detection of acute pneu­
monia was poor (31,71 %). However, the specificity and accuracy were
85.83 % and 72.62 %. Remarkably, the subgroup with none or slight
gravity-dependent attenuation performed better than subgroups with
advanced or severe gravity-dependent attenuation with an accuracy of
82,05 % versus 69.33 % and 70.37 %, respectively. Among patients who
underwent scanning within 16 h post-mortem, PMCT demonstrated
higher sensitivity (40.00 % vs. 18.75 %). However, its accuracy was
slightly lower compared to scans conducted after 16 h (71.08 % versus
74.12 %). Computed tomography identified 31 infiltrates suspected of
Fig. 2. Flowchart delineating the inclusion and exclusion of patients. pneumonia in 168 lobes. Eighteen infiltrates were false positive ac­
cording to histopathological analysis. Notably, congestion was detected
in 15 of the 18 lobes with a false positive infiltrate (six minimal, eight
Table 1 moderate and one severe). Also, the PMCT certainty of diagnosis was
Demographics table including patient characteristics, clinical parameters and probable in 13 false positives, and definite in five cases, suggesting that
prevalence of radiological findings. PTI= Postmortem time interval. false positives are less common when the certainty of diagnosis is higher.
Clinical and radiological features N = 53

Male/female 36/17 Discussion

Mean age (years) 66.7 (± 13)
Recent CPR performed 9 (17 %) Several studies have documented the challenges and limitations of
Last measured temperature within last 48 hours (◦ C) 37.0 (± 1.5) interpreting lung abnormalities encountered on PMCT imaging. Scan­
Last measured CRP (normal <10.0 mg/L) 136.7 (± 125.6)
Leucocyte count (3.5-11.0 10E9/L) 12.2 (± 8.24)
ning in an expiratory state and the presence of pulmonary edema make
Mean PTI (hh:mm) 22:47 (± 17:18) this interpretation specifically challenging. We hypothesized that the
PTI ≤ 16 hours 25 patients (47 %) severity of pulmonary edema is inversely proportional to the diagnostic
PTI > 16 hours 28 patients (53 %) value of PMCT in the detection of acute pneumonia. For this purpose,
Pleural effusion 20 (38 %)
PMCT scans and histological slices of 53 patients were re-examined and
Symmetrical/asymmetrical 8/12
Right-/left sided 17/14 analyzed. Additionally, the correlation between the PTI and severity of
Pre-existing lung disease 19 (36 %) gravity-dependent attenuation was assessed.
Individual lobes N = 212
Our results show a weak, yet statistically significant correlation be­
tween the PTI and the severity of gravity-dependent attenuation. This
Gravity-dependent attenuation
indicates that the duration of the postmortem interval (PTI) has a limited
​
Severity 0 (none) 7 (3 %)
Severity 1 (slight) 32 (15 %) influence on the severity of gravity-dependent attenuation in the lungs
Severity 2 (advanced) 75 (35 %) on PMCT. Additionally, it appears that the PTI is not the sole parameter
Severity 3 (severe) 54 (26 %) determining the extent of gravity-dependent attenuation severity. We
Severity 4 (fully opacified) 44 (21 %) did not find other parameters that explain the severity of gravity-
Certainty of the diagnosis acute pneumonia (n=168)
dependent attenuation in our analysis. The sensitivity of PMCT for the
​
Definitely not 43 (20 %)
Probably not 66 (31 %) detection of acute pneumonia was poor (31.71 %), illustrating the dif­
Uncertain 28 (13 %) ficulties with in identifying pulmonary pathology. Specificity was high
Probable 20 (9 %) (85.83 %) and accuracy acceptable (72.62 %). Diagnostic accuracy
Definite 11 (5 %)
decreased with increasing severity of gravity-dependent attenuation.
PTI < 16 h 83 (49 %
PTI > 16 h 85 (51 %) Sensitivity was higher in patients scanned within 16 h than in patients
scanned after 16 h. However, it is essential to interpret these results
cautiously, given the broad confidence intervals resulting from the low
Severity of gravity-dependent attenuation prevalence and constrained sample size in this retrospective study. Of
note, when bootstrapping was used as a resampling method to correct
The severity of gravity-dependent attenuation showed only weak for potential clustering (because multiple lobes were included per pa­
correlations with biochemical markers such as C-reactive protein tient), confidence intervals became even wider. The diagnostic param­
(-0.186, p = 0.007), and the leucocyte count (0.058, p = 0.413). Like­ eters after bootstrapping were not included in the results because
wise, no relevant correlation was found between the severity of gravity- bootstrapping could not be performed for all subgroups due to a limited
dependent attenuation and the presence of organizing pneumonia prevalence. Interestingly, a false positive diagnosis on PMCT nearly

4
M.G. Mentink et al. Forensic Imaging 40 (2025) 200617

Fig. 3. Boxplot of the PTI for gravity-dependent attenuation subgroups. PTI= Postmortem time interval.

acute pneumonia was the most frequently misdiagnosed condition

Table 2
among the cases with discrepancies, accounting for 10 out of 29 in­
Diagnostic measures of accuracy of PMCT for the detection of pneumonia,
stances. This underscores the particular significance of accurate diag­
including subgroup analyses. CI= Confidence interval.
nosis in cases involving this specific disease. In 2011, Shiotani et al.
Sensitivity Specificity PPV (95 NPV (95 Accuracy studied the paired data of three cadavers, reporting the progression of
(95 % CI) (95 % CI) % CI) % CI) (95 % CI)
gravity-dependent attenuation on the second PMCT [13]. Similarly,
Total lobes 31.71 % 85.83 % 41.94 % 79.56 % 72.62 % Hyodoh et al. analyzed the volume of pleural effusion in paired samples
(n = 168) (18.08 % to (78.53 % to (27.97 (75.75 (65.22 %
of 12 cadavers and reported a significant increase of volume over time
48.09 %) 91.38 %) % to % to to 79.20
57.32 82.91 %) until 40 h after death [14]. Lastly, Wagensveld et al. elaborately re­
%) %) ported on the appearance of postmortem changes in in-hospital deaths
Severity of gravity-dependent attenuation [5]. In their study the mean PTI was 23.0 h. Concerning the lungs, a
None + 60.00 % 89.66 % 66.67 % 86.67 % 82.05 % significant correlation was reported between the PTI and internal livores
slight (n = (26.24 % to (72.65 % to (37.95 (75.08 (66.47 %
39) 87.84 %) 97.81 %) % to to 93.34 to 92.46
of the lungs. While statistically significant the strength or weakness of
86.74 %) %) this correlation was not reported. Additionally, there was no significant
%) correlation found between the PTI and gravity-dependent changes (of
Advanced 22.73 % 88.68 % 45.45 % 73.44 % 69.33 % which the organ was not specified).
(n = 75) (7.82 % to (76.97 % to (22.10 (68.37 (57.62 %
To the best of our knowledge, this is the first time diagnostic pa­
45.37 %) 95.73 %) % to % to to 79.47
71.00 77.96 %) rameters for PCMT in the detection of acute pneumonia are presented in
%) %) literature. Establishing the diagnostic parameters of PMCT is of signifi­
Severe (n 22.22 % 80.00 % 18.18 % 83.72 % 70.37 % cant importance to radiologists. The low sensitivity and high specificity
= 54) (2.81 % to (65.40 % to (5.42 % (77.89 (56.39 % indicate that PMCT is more suitable to exclude acute pneumonia than to
60.01 %) 90.42 %) to 46.27 % to to 82.02)
%) 88.25
detect its presence. This is a key finding that could influence the radi­
%) ologist’s interpretation of the thorax in PMCT. In contrast to in-vivo CT
Postmortem time interval imaging where high diagnostic accuracy has been reported for
≤ 16 h (n = 40.00 % 84.48 % 52.63 % 76.56 % 71.08 % numerous pathologies, is seems that PMCT is limited for the detection of
83) (21.13 % to (72.58 % to (34.00 (69.96 (60.09 %
pulmonary pathology such as a pneumonia. The results also give support
61.33 %) 92.65 %) % to % to to 80.52
70.56 82.09 %) to the search for ventilation techniques to further improve the diagnostic
%) %) performance of PMCT.
> 16 h (n = 18.75 % 86.96 % 25.00 % 82.19 % 74.12 % The main limitation of this study is the study population size.
85) (4.05 % to (76.68 % to (9.22 % (78.19 (63.48 % Although the disease prevalence was comparable to published literature,
45.65 %) 93.86 %) to 52.24 % to to 83.01
%) 85.59 %)
the total number of lobes with acute pneumonia was small. Hence, we
%) underscore the importance of interpreting the results with caution. As a
result, confidence intervals were wide, and performing bootstrapping
methodology or conducting comparative tests between subgroups was
always showed the presence of congestion in the histopathological not meaningful. Although radiologists and pathologists were blinded to
analysis (indicative of antemortem pulmonary edema), and the certainty the original PMCT and autopsy reports, they were not blinded to clinical
of the radiologist was more often probable than definite of these and biochemical information. However, some clinical parameters, such
diagnoses. as temperature, and blood results are essential discriminators also used
Several other studies reported relevant findings that are in concor­ in daily clinical practice. Sampling errors are an inevitable occurrence in
dance with our results. First of all, Berezowska et al. have reported on histopathological analysis. It is uncertain to which degree and in what
the histopathological correlation of radiological findings. A similar manner sampling error has affected the results. Lastly, an important
prevalence of acute pneumonia is reported at 24 % [27]. Furthermore, limitation is the time interval between the time of the PMCT and the

5
M.G. Mentink et al. Forensic Imaging 40 (2025) 200617

autopsy, which was not analyzed. Although in daily routine autopsy observational cohort study in a Dutch tertiary referral centre, BMJ Open. 8 (2018)
e018834.
usually followed PMCT quickly, postmortem changes keep occurring in
[4] Y Kawasumi, A Usui, Y Hosokai, et al., Post-mortem computed tomography
this time window, creating a discrepancy between PMCT and autopsy findings of the lungs: Retrospective review and comparison with autopsy results of
findings. It is therefore (theoretically) possible that autopsy structurally 30 infant cases, Eur. J. Radiol. 84 (2015) 721–725.
reported more severe findings than were present at the time of PMCT [5] IM Wagensveld, BM Blokker, PA Wielopolski, et al., Total-body CT and MR features
of postmortem change in in-hospital deaths, PLoS. One 12 (2017) e0185115.
acquisition. [6] M Proisy, AJ Marchand, P Loget, et al., Whole-body post-mortem computed
In the future, it would be interesting to know whether diagnostic tomography compared with autopsy in the investigation of unexpected death in
accuracy improves when ventilated PMCT is performed. If so, it could infants and children, Eur. Radiol. 23 (2013) 1711–1719.
[7] IS Roberts, RE Benamore, EW Benbow, et al., Post-mortem imaging as an
justify its use in daily practice, because diagnosing acute pneumonia in alternative to autopsy in the diagnosis of adult deaths: a validation study, Lancet
in-hospital death would further improve postmortem diagnostics, 379 (2012) 136–142.
especially in cases where consent for autopsy is not provided. [8] MG Mentink, BGH Latten, FCH Bakers, et al., Efficacy of postmortem CT and tissue
sampling in establishing the cause of death in clinical practice: a prospective
observational study, J. Clin. Pathol. (2022).
Conclusion [9] PJ Robinson, L. Kreel, Pulmonary tissue attenuation with computed tomography:
comparison of inspiration and expiration scans, J. Comput. Assist. Tomogr. 3
(1979) 740–748.
Interpretation of pulmonary abnormalities encountered on PMCT [10] LJ Rosenblum, RA Mauceri, DE Wellenstein, et al., Density patterns in the normal
can pose considerable challenges, given the impact of postmortem lung as determined by computed tomography, Radiology. 137 (1980) 409–416.
changes and the acquisition of images in an expiratory state. The char­ [11] JA Verschakelen, L Van fraeyenhoven, G Laureys, M Demedts, AL. Baert,
Differences in CT density between dependent and nondependent portions of the
acteristic gravity-dependent attenuation appearance of pulmonary
lung: influence of lung volume, AJR Am. J. Roentgenol. 161 (1993) 713–717.
edema increases slightly as time progresses. The diagnostic performance [12] WR Webb, EJ Stern, N Kanth, G. Gamsu, Dynamic pulmonary CT: findings in
of PMCT for the detection of acute pneumonia was negatively affected healthy adult men, Radiology 186 (1993) 117–124.
by the severity of gravity-dependent attenuation and the PTI. The ra­ [13] S Shiotani, T Kobayashi, H Hayakawa, K Kikuchi, M. Kohno, Postmortem
pulmonary edema: a comparison between immediate and delayed postmortem
diologists should refrain from diagnosing acute pneumonia, but can rule computed tomography, Leg. Med. (Tokyo) 13 (2011) 151–155.
it out with an acceptable level of certainty. Future work could clarify the [14] H Hyodoh, J Shimizu, S Watanabe, S Okazaki, K Mizuo, H. Inoue, Time-related
added value of mechanical ventilation as it was shown that PMCT has a course of pleural space fluid collection and pulmonary aeration on postmortem
computed tomography (PMCT), Leg. Med. (Tokyo) 17 (2015) 221–225.
low sensitivity for this relevant diagnosis. [15] SH Durlacher, WG Banfield Jr., AD Bergner, Post-mortem pulmonary edema, Yale
J. Biol. Med. 22 (1950) 565–572.
CRediT authorship contribution statement [16] W Gonoi, Y Watanabe, G Shirota, et al., Pulmonary postmortem computed
tomography of bacterial pneumonia and pulmonary edema in patients following
non-traumatic in-hospital death, Leg. Med. (Tokyo) 45 (2020) 101716.
Max G. Mentink: Writing – original draft, Visualization, Validation, [17] S Shiotani, M Kohno, N Ohashi, et al., Non-traumatic postmortem computed
Software, Resources, Project administration, Methodology, Investiga­ tomographic (PMCT) findings of the lung, Forensic Sci. Int. 139 (2004) 39–48.
[18] L Filograna, MJ. Thali, Post-mortem CT imaging of the lungs: pathological versus
tion, Formal analysis, Data curation, Conceptualization. Bartholomeus non-pathological findings, Radiol. Med. 122 (2017) 902–908.
G.H. Latten: Writing – review & editing, Validation, Methodology, [19] M Ishida, W Gonoi, H Okuma, et al., Common postmortem computed tomography
Investigation, Data curation, Conceptualization. Frans C.H. Bakers: findings following atraumatic death: differentiation between normal postmortem
changes and pathologic lesions, Korean J. Radiol. 16 (2015) 798–809.
Writing – review & editing, Methodology, Investigation, Data curation.
[20] C Robinson, MJ Biggs, J Amoroso, M Pakkal, B Morgan, GN. Rutty, Post-mortem
Casper Mihl: Writing – review & editing, Methodology, Investigation, computed tomography ventilation; simulating breath holding, Int. J. Legal. Med.
Data curation, Conceptualization. Lisa M. Hillen: Writing – review & 128 (2014) 139–146.
editing, Investigation, Data curation, Conceptualization. Paul A.M. [21] GN Rutty, B Morgan, T Germerott, M Thali, O. Athurs, Ventilated post-mortem
computed tomography – a historical review, J. Forensic Radiol. Imaging 4 (2016)
Hofman: Writing – review & editing, Supervision, Methodology, 35–42.
Investigation, Conceptualization. [22] A Christe, P Flach, S Ross, et al., Clinical radiology and postmortem imaging
(Virtopsy) are not the same: Specific and unspecific postmortem signs, Leg. Med.
(Tokyo) 12 (2010) 215–222.
Declaration of competing interest [23] T Germerott, PM Flach, US Preiss, SG Ross, MJ. Thali, Postmortem ventilation: a
new method for improved detection of pulmonary pathologies in forensic imaging,
The authors declare that they have no known competing financial Leg. Med. (Tokyo) 14 (2012) 223–228.
[24] GN Rutty, MJ Biggs, A Brough, et al., Ventilated post-mortem computed
interests or personal relationships that could have appeared to influence tomography through the use of a definitive airway, Int. J. Legal. Med. 129 (2015)
the work reported in this paper. 325–334.
[25] MG Mentink, FCH Bakers, C Mihl, et al., Introduction of postmortem CT increases
the postmortem examination rate without negatively impacting the rate of
References traditional autopsy in daily practice: an implementation study, J. Clin. Pathol. 74
(2020) 177–181.
[1] SE Westphal, J Apitzsch, T Penzkofer, AH Mahnken, R. Knuchel, Virtual CT autopsy [26] MG Mentink, BGH Latten, FCH Bakers, et al., Clinical relevance of unexpected
in clinical pathology: feasibility in clinical autopsies, Virchows Arch. 461 (2012) findings of post-mortem computed tomography in hospitalized patients: an
211–219. observational study, Int. J. Environ. Res. Public Health 17 (2020) 7572.
[2] K Inai, S Noriki, K Kinoshita, et al., Postmortem CT is more accurate than clinical [27] S Berezowska, A Schmid, T Losmanová, et al., Frequency and significance of
diagnosis for identifying the immediate cause of death in hospitalized patients: a pathologic pulmonary findings in postmortem examinations-a single center
prospective autopsy-based study, Virchows Arch. 469 (2016) 101–109. experience before COVID-19, Diagnostics. (Basel) (2021) 11.
[3] LJP Sonnemans, B Kubat, M Prokop, WM. Klein, Can virtual autopsy with
postmortem CT improve clinical diagnosis of cause of death? A retrospective