Jpn J Radiol DOI 10.1007/s11604-014-0326-9
ORIGINAL ARTICLE
Postmortem lung features in drowning cases on computed tomography Akihito Usui • Yusuke Kawasumi Masato Funayama • Haruo Saito
•
Received: 13 February 2014 / Accepted: 24 April 2014 Ó Japan Radiological Society 2014
Abstract Purpose We sought to compare postmortem chest computed tomography (CT) features of drowning cases with autopsy findings, and to classify these features. Materials and method We performed a retrospective analysis of high-resolution and multi-planar reconstruction chest CT images of drowning in 92 adults (54 men, 38 women; mean age 65.4 years) scanned before forensic autopsy. The average lung CT number was calculated from whole-lung images reconstructed on a 3D workstation. The statistically significant differences of CT numbers were assessed with an alpha level of 0.05. Results Postmortem chest CT image patterns were classified into six types: the two main types were ground-glass opacities with thickened pulmonary interstitium (n = 31), and a centrilobular distribution of ill-defined nodules along the airways (n = 38). Some cases were mixed type (n = 10). There were significant differences in CT numbers between each type. The remaining three types were consolidation (n = 5), emphysema and/or fibrosis (n = 4), and unclassifiable (n = 4).
A. Usui (&) H. Saito Department of Diagnostic Image Analysis, Tohoku University Graduate School of Medicine, 2-1 Seiryo-machi, Sendai 980-8575, Japan e-mail: [email protected] Y. Kawasumi Department of Clinical Imaging, Tohoku University Graduate School of Medicine, 2-1 Seiryo-machi, Sendai 980-8575, Japan M. Funayama Department of Forensic Medicine, Tohoku University Graduate School of Medicine, 2-1 Seiryo-machi, Sendai 980-8575, Japan
Conclusion Postmortem CT images of drowning cases can be classified into three major types with a few exceptions. Keywords Drowning Lung Postmortem computed tomography Postmortem examination Forensic radiography
Introduction According to the annual World Health Organization report 2012 [1], *390,000 persons worldwide die annually as a result of drowning. In 2013, vital statistics on Japan from the Japanese Health, Labor, and Welfare Ministry reported that the numbers of accidental and suicidal deaths by drowning were 7,963 and 1,168, respectively. Although laypeople may believe that it is easy to ascribe a death to drowning, from the medicolegal viewpoint, drowning is often called a diagnosis of exclusion [2]. When a fresh body is found in water, the diagnosis of drowning is based on circumstances surrounding the death and some nonspecific autopsy findings that are generally known to be observed in evidential drowning cases [3]. When a victim inhales a sizable volume of water, some findings are seen in the airways, e.g., white or bloody foam is present in the mouth and airways, and the lungs are bulky because of edema and overexpansion [2–4]. However, these findings are nonspecific, as they can be observed in victims dying of drug overdose, carbon monoxide poisoning, or other causes of death. Recently, computed tomography (CT) has been used to obtain imaging data in forensic cases [5, 6]. Over the past few years, more than 20 forensic institutes in the 90 Japanese schools of medicine have acquired CT scanners for
123
Jpn J Radiol
use in forensic autopsy cases. In drowning death, some CT findings are felt to be useful. One is fluid accumulation due to water inflow into the paranasal sinuses [7]. On routine autopsy, it is hard to check the state of the paranasal sinuses, but CT images can give reliable information noninvasively. Also useful are the pulmonary findings. Macroscopically, the appearance of the cut surface of the drowned lung may be light red in color with small amounts of blood and large quantities of edema fluid. Unfortunately, macroscopic gross observation has limitations in the evaluation of pulmonary edema. Many pathologists have experienced several cases with only mild pulmonary edema despite obvious drowning. CT is quite accurate in quantifying edema. Some researchers have stated that groundglass opacity (GGO) involving the central part of the secondary pulmonary lobule and/or centrilobular nodules spreading out and surrounding the airway indicate aspiration of water [8]. Another study has also reported findings of GGO with septal lines in drowned lungs on postmortem CT [9, 10]. There has been no classification of pulmonary edema types using postmortem CT in drowning cases. In some clinical cases, plain chest radiographs have been used to confirm the diffuse pulmonary edema produced by aspiration of massive amounts of water [11, 12]. In an emergency setting, there is no time for chest CT scanning, resulting in a paucity of reports of CT of near-drowning cases in the emergency medicine literature. Shiotani et al. [13] reviewed postmortem CT findings of the lung in many cases of various non-traumatic deaths; however, there were no cases of drowning. In this study, we evaluated the features of drowning death cases on postmortem chest CT and aimed to establish a classification system.
Materials and methods We performed a retrospective CT analysis of drowning cases in 92 adults (54 men, 38 women; age range 44–90 years; mean age 65.4 years) scanned from May 2009 to September 2013 with multi-slice CT (MSCT) before forensic autopsy. The numbers of bodies found in salt water (e.g., sea, seacoast, and estuarine) and fresh water (e.g., river, lake, rice field, water well, irrigation ditch, and pond) were 33 and 59, respectively. The forensic diagnosis of drowning was based on the investigation of the death scene by the police and on medicolegal autopsies, including drug screens and diatom tests [2, 3, 14, 15]. In this study, to exclude putrefactive artifacts, we analyzed scans performed no later than two days after death. Cases of bathtub death were also excluded from the study, because pathological factors such as myocardial infarction may largely contribute to the cause of death in this setting [13]. The use of postmortem CT images for this research
123
was approved by the ethics board of our institution. Informed consent was not required for this research. CT scanners and imaging In our institution, postmortem MSCT scanning as part of preautopsy screening was performed on an eight-channel scanner (Aquilion: Toshiba Medical Systems, Tokyo, Japan). Each corpse was covered with three blankets and sealed in an opaque body sheet at the autopsy suite to avoid infectious contamination during MSCT scanning. Before April 2012, we had performed helical scanning with 2.0 mm collimation. Subsequently, we obtained axial conventional scan images of the chest in addition to the helical data. The present protocol of postmortem MSCT scanning is as follows. High-resolution (HR) chest scans were obtained with 135 kVp, 190–250 mAs, M-sized field of view (FOV), and 1.0 mm slices every 25–30 mm through the thorax in four-row multislice mode and processed with lung kernel settings. Following the lung HR-CT scanning, volumetric helical scans were obtained from the head to the proximal femurs at 120 kVp with variable mAs, a beam pitch of 0.875, LL-sized FOV, and 2.0-mm collimation. The volumetric data allowed reconstruction of whole lung images from 2.0-mm slices, also with lung kernel settings. All postmortem radiographic data were sent to a digital imaging and communications in medicine (DICOM) server (POP-Net Server; ImageONE, Tokyo, Japan) and a three-dimensional (3D) workstation (Ziostation ver. 2.1.5.0; Ziosoft, Tokyo, Japan). Assessment of lung images We assessed the HR-CT images and multiplanar reconstruction (MPR) images provided by the helical scanning data and classified postmortem CT features of the lungs. Two board certified radiologists with more than 10 and 25 years of experience, respectively, read the lung images. These radiologists also have more than 4 years of experience in interpreting postmortem CT. The averaged bilateral whole lung CT number was calculated on a 3D workstation, as shown in Fig. 1, using 3D reconstructions. Statistical analysis The Tukey–Kramer method was used to compare categorical variables as the differences in CT numbers. Significance was determined when the p value was \0.05.
Results We classified postmortem CT findings of the lungs in drowning into two main types: Type 1, showing diffuse
Jpn J Radiol Fig. 1 Screen image of threedimensional (3D) workstation for calculation of the whole lung computed tomography (CT) number. The three images on the left side show axial (from cephalad to caudad), coronal, and sagittal images. These are the selected lung areas for 3D image reconstruction, as shown in the right-side image. The average CT number of the total voxels in the 3D image is calculated
GGO with thickening of the acinar and lobular interstitium (Fig. 2), and Type 2, showing ill-defined centrilobular nodules and patchy GGO spreading along the airways (Fig. 3). Some cases showed combinations of Type 1 and Type 2. We designated them as ‘‘Type 1 ? 2’’ (Fig. 4). In 79 out of 92 cases, the numbers of Type 1, Type 2, and Type 1 ? 2 were 31 (34 %), 38 (41 %), and 10 (11 %), respectively. At autopsy, severe edema was observed in 97 % (29 of 31) of Type 1 cases. In these cases, the cut lung surface had a bulging light brick-red appearance, with prominent expression of red transparent fluid and white and/or bloody foamy fluid (Fig. 2). In the other two Type 1 cases, the amount of edema was moderate. Type 2 cases were characterized by a similar reddish, bulging cut surface, but the amount of edema ranged from severe (n = 16) (Fig. 3) to moderate (n = 17) to mild (n = 5). In cases of Type 1 ? 2, like Type 2, the amount of pulmonary edema varied [severe n = 6 (Fig. 4), moderate n = 3, and mild n = 1]. The remaining types were Type 3: consolidation (n = 5) (Fig. 5), Type 4: emphysema and/or fibrosis (n = 4) (Fig. 6), and unclassifiable (n = 4) (Fig. 7). All Type 3 cases had undergone cardiopulmonary resuscitation, including fluid replacement, at an emergency department for an unknown period of time. All Type 4 cases had typical findings of pulmonary fibrosis, including distributed tiny air cysts. This was accompanied by prominent bullous changes in two of the four cases.
Among the four cases of unclassifiable type, one is shown in Fig. 7. Two of these cases were found at the bottom of a river. The lungs had widespread geographic GGO with lobular distribution forming a mosaic pattern. The autopsy findings revealed severe pulmonary edema with frothy fluid transudate and the presence of diatoms in the periphery of the lungs. In another case, the body was found in a prone position with the face immersed under the surface of an 8-cm-deep irrigation ditch. This case had slightly increased opacity involving both lungs. At autopsy, foamy fluid was found in the airway, and there was mild pulmonary edema. The diatom test was positive. In the fourth case (Fig. 7), the body was found in the right lateral decubitus position in an irrigation ditch, with the right half of the body and the nasal cavity completely underwater when found. Relatively severe pulmonary edema was noted on the cut surface of the left upper lobe and mild pulmonary edema was present in the other lung lobes. The diatom test was positive. It was difficult to find similarities between these cases and Types 1, 2, or 1 ? 2. Table 1 demonstrates the average CT numbers, which are in decreasing order, of Types 1, 1 ? 2, and 2. Differences between each type were significant (Fig. 8). For Type 3, Type 4, and the unclassifiable type, we could not perform a test of statistical significance with the Tukey–Kramer method because of the small number of cases.
123
Jpn J Radiol
Fig. 2 Postmortem CT image of lungs in drowning classified as Type 1 (a). Ground-glass opacity (GGO) with a combination of interlobular septal thickening and interstitial thickening spreads diffusely. Cut section of both lungs (b) shows a light brick-red color and a bulging cut surface, and large quantities of red transparent edema fluid (asterisks) and white foam flowing out from each cut surface and bronchi (arrows). The right and left lungs weighed 830 and 750 g, respectively
Discussion
Fig. 3 Postmortem CT image of lungs in drowning classified as Type 2 (a). Centrilobular nodules spread out surrounding the airways without interlobular septal thickening (white arrows). Bilateral pleural effusions are also present. This case was macroscopically judged as severe lung edema by the forensic pathologist. Cut sections of both lungs show a light reddish appearance similar to that in Fig. 2b except for white foam. The right (b) and left (c) lungs weighed 550 and 480 g, respectively
The lung CT numbers of drowning cases are mainly influenced by the proportion of inhaled water in the normal lobular lung parenchyma [8]. This means that the airspaces in Type 1 were filled with more water than those of Type 2. Macroscopically, forensic pathologists classify postmortem pulmonary edema as severe, moderate, or mild. The lungs with Type 1 CT findings were usually classified as severe edema. However, the lungs with Type 2 CT findings were classified with a wide range of levels of edema. It is difficult to evaluate the severity of pulmonary edema using macroscopic examination of the lung cut surface in a standard
fashion, because visual assessment is so subjective. Postmortem CT may provide a more objective assessment of pulmonary edema, as the findings are fairly specific. Many forensic textbooks show that one of the characteristic pulmonary findings seen in drowning is called emphysema aquosum [2, 3, 16]. This term is used to describe hyperexpanded and waterlogged lungs with no putrefactive change, implying that the lungs are markedly overinflated, and the findings are the result of prominent expansion by inhaling a combination of air and water while
123
Jpn J Radiol Fig. 4 Postmortem CT image of lungs in drowning classified as Type 1 ? 2 (a). Centrilobular nodules spread out surrounding the airways in the upper lobes (black arrows), and GGO with a superimposed reticular pattern consisting of a combination of interlobular septal thickening and interstitial thickening is widespread in the lower lobes (black arrowhead). The former and latter signs are often observed in Type 2 and Type 1, respectively. This case was macroscopically judged as severe lung edema by the forensic pathologist. Cut sections of the right (b) and left lungs (c) show a bulging cut surface with a brick-red appearance and white foam flowing out similar to Fig. 2b. The right and left lungs weighed 730 and 600 g, respectively
drowning [2, 3, 14, 16]. Previous studies have reported that postmortem ventilation leads to a clearly visible decrease in GGO due to an increase in air in the interstitial space [17, 18]. Thus, a large volume of inhaled/instilled air in the airway, producing a hyperexpanded lung that is the characteristic finding of drowning, should be accompanied by decreased lung CT numbers, and could explain the appearance of faint opacification in Type 2. We believe that emphysema aquosum might be the reason for the Type 2 appearance. The CT appearance of Type 1 is diffuse GGO with thickening of the pulmonary interstitium. The GGO was recognized as two phenomena that often occur simultaneously as observed within the resolution of CT; that is, a mild decrease in the amount of air in the airspaces and a mild increase in fluid content, with preserved visibility of the bronchial and vascular markings because they are outlined by residual alveolar air [8, 19, 20]. Minimal thickening of the alveolar walls and of the parenchymal
interstitium can also cause GGO [8]. In Type 1, the reticular pattern resulting from thickened interlobular septa mainly results from the infiltration of the interstitium with fluid, reflecting pulmonary edema [8, 19, 21]. On the other hand, Type 2 demonstrates faint centrilobular nodules and patchy opacities that may be formed by the fusion of nodules distributed along the airways. The terminal bronchioles are located in the central core of the secondary pulmonary lobules [8, 21]; some alveolar spaces filled with fluid may give this appearance. Therefore, CT number measurements of these two types of drowning lung may be able to distinguish between water in the airways (Type 2) and water in the interstitium (Type 1). Figure 4 shows a case of Type 1 ? 2, meaning that centrilobular nodules surround the airways in both upper lobes and reticular interstitial patterns involve both lower lobes. In this case, according to the police investigation, the body was found floating prone in a river. At autopsy, a large amount of white frothy fluid was found in the airway, with severe
123
Jpn J Radiol
Fig. 5 Postmortem CT image of lungs classified as consolidation Type 3. Irregular consolidation is widespread
Fig. 7 Postmortem CT image of lungs of the unclassified type. The image shows isolated hazy opacities in the right upper lobe (white arrow) and band-like GGO in the left upper lobe (white arrowhead). The other lung parenchyma appears normal
Table 1 Classification table of type of lung findings on postmortem computed tomography (CT) Type of lung findings
Number of cases
CT number ± SD/HU
Type 1
31 (34 %)
-471 ± 85
Type 2
38 (41 %)
-695 ± 66
Type 1 ? 2
10 (11 %)
-606 ± 85
Type 3 (consolidation)
5 (5 %)
-242 ± 162
Type 4 (emphysema/fibrosis)
4 (4 %)
-591 ± 41
Unclassifiable
4 (4 %)
-551 ± 200
Total
92
SD standard deviation, HU Hounsfield unit
Fig. 6 Postmortem CT image of lungs in drowning classified as emphysema and/or fibrosis type. Bullous emphysema and interstitial fibrosis are observed in the lungs. Massive bilateral pleural effusions are also seen
pulmonary edema and a moderate number of diatoms in the lung periphery. Some possible explanations for Type 1 ? 2 may be postmortem change, a transition from Type 2 to Type 1, independent mechanisms in the different lobes, and the intermingling of them. We think further investigation is needed to elucidate the etiology of Type 1 ? 2.
123
Consolidation is defined as an increase in lung attenuation with obscuration of the underlying vessels [8], indicating a decrease in the amount of air in the airspaces and replacement by fluid or other substances. The five cases of consolidation (Type 3) had undergone transfusion during the resuscitation procedure. The autopsy findings of the lungs demonstrated marked edema with little aeration. In this report, we classified it as a special type related to drowning, but to conclude, it may be nonspecific edema due to transfusion therapy. The emphysema and/or fibrosis of Type 4 was histologically confirmed. The lungs did not show the typical findings of drowning, although the death scene was demonstrative. Thus, the CT findings suggested preexisting
Jpn J Radiol
References
Fig. 8 Box plot of the CT number distribution by three Types of CT appearance. The median values as drawn lines in the boxes of each Type (Type 1, Type 1 ? 2, and Type 2) were -492, -624, and -709 HU, respectively. The top and bottom of the boxes indicate the upper and lower quartiles, respectively. Significant differences were observed between each CT pattern Type by the Tukey–Kramer method
pulmonary emphysema and fibrosis, but were unhelpful in diagnosing drowning. When a body is found submerged in water, one of the most important questions is whether the victim inhaled water or not. The autopsy rate in Japan is very low compared with many other developed countries. Many bodies are not autopsied and the cause of death is decided only by external examination. In this article, we report the CT images of drowning cases mainly classified into three main types and two minor types, with a few unclassifiable. This study shows that postmortem CT images can provide significant information about drowning. When one uses postmortem lung CT images to diagnose whether the decedent died by drowning, the remaining issue is the existence of unclassified cases. In this study, unclassified cases amounted to only 4.3 % of total cases. However, when postmortem lung CT shows some image patterns other than the five types, death by drowning cannot be excluded. To clarify why these unique images appear in drowning cases, further comparisons between postmortem lung CT images and autopsy findings are necessary. Conflict of interest disclosed.
No potential conflicts of interest were
1. World Health Organization Fact sheet N347. Drowning. Geneva, Switzerland: WHO; 2012. 2. Spitz DJ. Investigation of bodies in water. In: Spitz WU, editor. Medico-legal investigation of death: guidelines for the application of pathology to crime investigation. 4th ed. Spring-field: Charles C Thomas; 2006. p. 846–81. 3. Saukko P, Knight B. Immersion deaths. In: Saukko P, Knight B, editors. Knight’s forensic pathology. 3rd ed. London: Edward Arnold; 2004. p. 395–411. 4. Christe A, Aghayev E, Jackowski C, Thali MJ, Vock P. Drowning-post-mortem imaging findings by computed tomography. Eur Radiol. 2008;18:283–90. 5. Persson A, Lindblom M, Jackowski C. A state-of-the-art pipeline for postmortem CT and MRI visualization: from data acquisition to interactive image interpretation at autopsy. Acta Radiol. 2011;52:522–36. 6. O’Donnell C, Woodford N. Post-mortem radiology––a new subspeciality? Clin Radiol. 2008;63:1189–94. 7. Kawasumi Y, Kawabata T, Sugai Y, Usui A, Hosokai Y, Sato M, et al. Assessment of the relationship between drowning and fluid accumulation in the paranasal sinuses on post-mortem computed tomography. Eur J Radiol. 2012;81:3953–5. 8. Verschakelen JA, De Wever W. Computed tomography of the lung. A pattern approach. Berlin: Springer; 2007. 9. Levy AD, Harcke HT, Getz JM, et al. Virtual autopsy: two- and three-dimensional multidetector CT findings in drowning with autopsy comparison. Radiology. 2007;243:862–8. 10. Kim KI, Lee KN, Tomiyama N, Johkoh T, Ichikado K, Kim CW, et al. Near drowning: thin-section CT findings in six patients. J Comput Assist Tomogr. 2000;24:562–6. 11. Hunter TB, Whitehouse WM. Fresh-water near-drowning: radiological aspects. Radiology. 1974;112:51–6. 12. Franquet T, Gime´nez A, Roso´n N, Torrubia S, Sabate´ JM, Pe´rez C. Aspiration diseases: findings, pitfalls, and differential diagnosis. Radiographics. 2000;20:673–85. 13. Shiotani S, Kohno M, Ohashi N, Yamazaki K, Nakayama H, Watanabe K, et al. Non-traumatic postmortem computed tomographic (PMCT) findings of the lung. Forensic Sci Int. 2004;139:39–48. 14. DiMaio VJ, DiMaio D. Forensic pathology. 2nd ed. Boca Raton: CRC Press; 2001. 15. Funayama M, Aoki Y, Sebetan IM, Sagisaka K. Detection of diatoms in blood by a combination of membrane filtering and chemical digestion. Forensic Sci Int. 1987;34:175–82. 16. Lunetta P, Modell JH. Macroscopical, microscopical, and laboratory findings in drowning victims: a comprehensive review. In: Tsokos M, editor. Forensic pathology reviews, vol. 3. Totowa: Humana Press; 2005. p. 3–77. 17. Germerott T, Preiss US, Ebert LC, Ruder TD, Ross S, Flach PM, et al. A new approach in virtopsy: postmortem ventilation in multislice computed tomography. Leg Med (Tokyo). 2010;12:276–9. 18. Germerott T, Flach PM, Preiss US, Ross SG, Thali MJ. Postmortem ventilation: a new method for improved detection of pulmonary pathologies in forensic imaging. Leg Med (Tokyo). 2012;14:223–8. 19. Storto ML, Kee ST, Golden JA, Webb WR. Hydrostatic pulmonary edema: high-resolution CT findings. Am J Roentgenol. 1995;165:817–20. 20. Gruden JF, Webb WR, Warnock M. Centrilobular opacities in the lung on high-resolution CT: diagnostic considerations and pathologic correlation. Am J Roentgenol. 1994;162:569–74. 21. Webb WR. Thin-section CT of the secondary pulmonary lobule: anatomy and the image––the 2004 Fleischner lecture. Radiology. 2006;239:322–38.
123
ORIGINAL ARTICLE
Postmortem lung features in drowning cases on computed tomography Akihito Usui • Yusuke Kawasumi Masato Funayama • Haruo Saito
•
Received: 13 February 2014 / Accepted: 24 April 2014 Ó Japan Radiological Society 2014
Abstract Purpose We sought to compare postmortem chest computed tomography (CT) features of drowning cases with autopsy findings, and to classify these features. Materials and method We performed a retrospective analysis of high-resolution and multi-planar reconstruction chest CT images of drowning in 92 adults (54 men, 38 women; mean age 65.4 years) scanned before forensic autopsy. The average lung CT number was calculated from whole-lung images reconstructed on a 3D workstation. The statistically significant differences of CT numbers were assessed with an alpha level of 0.05. Results Postmortem chest CT image patterns were classified into six types: the two main types were ground-glass opacities with thickened pulmonary interstitium (n = 31), and a centrilobular distribution of ill-defined nodules along the airways (n = 38). Some cases were mixed type (n = 10). There were significant differences in CT numbers between each type. The remaining three types were consolidation (n = 5), emphysema and/or fibrosis (n = 4), and unclassifiable (n = 4).
A. Usui (&) H. Saito Department of Diagnostic Image Analysis, Tohoku University Graduate School of Medicine, 2-1 Seiryo-machi, Sendai 980-8575, Japan e-mail: [email protected] Y. Kawasumi Department of Clinical Imaging, Tohoku University Graduate School of Medicine, 2-1 Seiryo-machi, Sendai 980-8575, Japan M. Funayama Department of Forensic Medicine, Tohoku University Graduate School of Medicine, 2-1 Seiryo-machi, Sendai 980-8575, Japan
Conclusion Postmortem CT images of drowning cases can be classified into three major types with a few exceptions. Keywords Drowning Lung Postmortem computed tomography Postmortem examination Forensic radiography
Introduction According to the annual World Health Organization report 2012 [1], *390,000 persons worldwide die annually as a result of drowning. In 2013, vital statistics on Japan from the Japanese Health, Labor, and Welfare Ministry reported that the numbers of accidental and suicidal deaths by drowning were 7,963 and 1,168, respectively. Although laypeople may believe that it is easy to ascribe a death to drowning, from the medicolegal viewpoint, drowning is often called a diagnosis of exclusion [2]. When a fresh body is found in water, the diagnosis of drowning is based on circumstances surrounding the death and some nonspecific autopsy findings that are generally known to be observed in evidential drowning cases [3]. When a victim inhales a sizable volume of water, some findings are seen in the airways, e.g., white or bloody foam is present in the mouth and airways, and the lungs are bulky because of edema and overexpansion [2–4]. However, these findings are nonspecific, as they can be observed in victims dying of drug overdose, carbon monoxide poisoning, or other causes of death. Recently, computed tomography (CT) has been used to obtain imaging data in forensic cases [5, 6]. Over the past few years, more than 20 forensic institutes in the 90 Japanese schools of medicine have acquired CT scanners for
123
Jpn J Radiol
use in forensic autopsy cases. In drowning death, some CT findings are felt to be useful. One is fluid accumulation due to water inflow into the paranasal sinuses [7]. On routine autopsy, it is hard to check the state of the paranasal sinuses, but CT images can give reliable information noninvasively. Also useful are the pulmonary findings. Macroscopically, the appearance of the cut surface of the drowned lung may be light red in color with small amounts of blood and large quantities of edema fluid. Unfortunately, macroscopic gross observation has limitations in the evaluation of pulmonary edema. Many pathologists have experienced several cases with only mild pulmonary edema despite obvious drowning. CT is quite accurate in quantifying edema. Some researchers have stated that groundglass opacity (GGO) involving the central part of the secondary pulmonary lobule and/or centrilobular nodules spreading out and surrounding the airway indicate aspiration of water [8]. Another study has also reported findings of GGO with septal lines in drowned lungs on postmortem CT [9, 10]. There has been no classification of pulmonary edema types using postmortem CT in drowning cases. In some clinical cases, plain chest radiographs have been used to confirm the diffuse pulmonary edema produced by aspiration of massive amounts of water [11, 12]. In an emergency setting, there is no time for chest CT scanning, resulting in a paucity of reports of CT of near-drowning cases in the emergency medicine literature. Shiotani et al. [13] reviewed postmortem CT findings of the lung in many cases of various non-traumatic deaths; however, there were no cases of drowning. In this study, we evaluated the features of drowning death cases on postmortem chest CT and aimed to establish a classification system.
Materials and methods We performed a retrospective CT analysis of drowning cases in 92 adults (54 men, 38 women; age range 44–90 years; mean age 65.4 years) scanned from May 2009 to September 2013 with multi-slice CT (MSCT) before forensic autopsy. The numbers of bodies found in salt water (e.g., sea, seacoast, and estuarine) and fresh water (e.g., river, lake, rice field, water well, irrigation ditch, and pond) were 33 and 59, respectively. The forensic diagnosis of drowning was based on the investigation of the death scene by the police and on medicolegal autopsies, including drug screens and diatom tests [2, 3, 14, 15]. In this study, to exclude putrefactive artifacts, we analyzed scans performed no later than two days after death. Cases of bathtub death were also excluded from the study, because pathological factors such as myocardial infarction may largely contribute to the cause of death in this setting [13]. The use of postmortem CT images for this research
123
was approved by the ethics board of our institution. Informed consent was not required for this research. CT scanners and imaging In our institution, postmortem MSCT scanning as part of preautopsy screening was performed on an eight-channel scanner (Aquilion: Toshiba Medical Systems, Tokyo, Japan). Each corpse was covered with three blankets and sealed in an opaque body sheet at the autopsy suite to avoid infectious contamination during MSCT scanning. Before April 2012, we had performed helical scanning with 2.0 mm collimation. Subsequently, we obtained axial conventional scan images of the chest in addition to the helical data. The present protocol of postmortem MSCT scanning is as follows. High-resolution (HR) chest scans were obtained with 135 kVp, 190–250 mAs, M-sized field of view (FOV), and 1.0 mm slices every 25–30 mm through the thorax in four-row multislice mode and processed with lung kernel settings. Following the lung HR-CT scanning, volumetric helical scans were obtained from the head to the proximal femurs at 120 kVp with variable mAs, a beam pitch of 0.875, LL-sized FOV, and 2.0-mm collimation. The volumetric data allowed reconstruction of whole lung images from 2.0-mm slices, also with lung kernel settings. All postmortem radiographic data were sent to a digital imaging and communications in medicine (DICOM) server (POP-Net Server; ImageONE, Tokyo, Japan) and a three-dimensional (3D) workstation (Ziostation ver. 2.1.5.0; Ziosoft, Tokyo, Japan). Assessment of lung images We assessed the HR-CT images and multiplanar reconstruction (MPR) images provided by the helical scanning data and classified postmortem CT features of the lungs. Two board certified radiologists with more than 10 and 25 years of experience, respectively, read the lung images. These radiologists also have more than 4 years of experience in interpreting postmortem CT. The averaged bilateral whole lung CT number was calculated on a 3D workstation, as shown in Fig. 1, using 3D reconstructions. Statistical analysis The Tukey–Kramer method was used to compare categorical variables as the differences in CT numbers. Significance was determined when the p value was \0.05.
Results We classified postmortem CT findings of the lungs in drowning into two main types: Type 1, showing diffuse
Jpn J Radiol Fig. 1 Screen image of threedimensional (3D) workstation for calculation of the whole lung computed tomography (CT) number. The three images on the left side show axial (from cephalad to caudad), coronal, and sagittal images. These are the selected lung areas for 3D image reconstruction, as shown in the right-side image. The average CT number of the total voxels in the 3D image is calculated
GGO with thickening of the acinar and lobular interstitium (Fig. 2), and Type 2, showing ill-defined centrilobular nodules and patchy GGO spreading along the airways (Fig. 3). Some cases showed combinations of Type 1 and Type 2. We designated them as ‘‘Type 1 ? 2’’ (Fig. 4). In 79 out of 92 cases, the numbers of Type 1, Type 2, and Type 1 ? 2 were 31 (34 %), 38 (41 %), and 10 (11 %), respectively. At autopsy, severe edema was observed in 97 % (29 of 31) of Type 1 cases. In these cases, the cut lung surface had a bulging light brick-red appearance, with prominent expression of red transparent fluid and white and/or bloody foamy fluid (Fig. 2). In the other two Type 1 cases, the amount of edema was moderate. Type 2 cases were characterized by a similar reddish, bulging cut surface, but the amount of edema ranged from severe (n = 16) (Fig. 3) to moderate (n = 17) to mild (n = 5). In cases of Type 1 ? 2, like Type 2, the amount of pulmonary edema varied [severe n = 6 (Fig. 4), moderate n = 3, and mild n = 1]. The remaining types were Type 3: consolidation (n = 5) (Fig. 5), Type 4: emphysema and/or fibrosis (n = 4) (Fig. 6), and unclassifiable (n = 4) (Fig. 7). All Type 3 cases had undergone cardiopulmonary resuscitation, including fluid replacement, at an emergency department for an unknown period of time. All Type 4 cases had typical findings of pulmonary fibrosis, including distributed tiny air cysts. This was accompanied by prominent bullous changes in two of the four cases.
Among the four cases of unclassifiable type, one is shown in Fig. 7. Two of these cases were found at the bottom of a river. The lungs had widespread geographic GGO with lobular distribution forming a mosaic pattern. The autopsy findings revealed severe pulmonary edema with frothy fluid transudate and the presence of diatoms in the periphery of the lungs. In another case, the body was found in a prone position with the face immersed under the surface of an 8-cm-deep irrigation ditch. This case had slightly increased opacity involving both lungs. At autopsy, foamy fluid was found in the airway, and there was mild pulmonary edema. The diatom test was positive. In the fourth case (Fig. 7), the body was found in the right lateral decubitus position in an irrigation ditch, with the right half of the body and the nasal cavity completely underwater when found. Relatively severe pulmonary edema was noted on the cut surface of the left upper lobe and mild pulmonary edema was present in the other lung lobes. The diatom test was positive. It was difficult to find similarities between these cases and Types 1, 2, or 1 ? 2. Table 1 demonstrates the average CT numbers, which are in decreasing order, of Types 1, 1 ? 2, and 2. Differences between each type were significant (Fig. 8). For Type 3, Type 4, and the unclassifiable type, we could not perform a test of statistical significance with the Tukey–Kramer method because of the small number of cases.
123
Jpn J Radiol
Fig. 2 Postmortem CT image of lungs in drowning classified as Type 1 (a). Ground-glass opacity (GGO) with a combination of interlobular septal thickening and interstitial thickening spreads diffusely. Cut section of both lungs (b) shows a light brick-red color and a bulging cut surface, and large quantities of red transparent edema fluid (asterisks) and white foam flowing out from each cut surface and bronchi (arrows). The right and left lungs weighed 830 and 750 g, respectively
Discussion
Fig. 3 Postmortem CT image of lungs in drowning classified as Type 2 (a). Centrilobular nodules spread out surrounding the airways without interlobular septal thickening (white arrows). Bilateral pleural effusions are also present. This case was macroscopically judged as severe lung edema by the forensic pathologist. Cut sections of both lungs show a light reddish appearance similar to that in Fig. 2b except for white foam. The right (b) and left (c) lungs weighed 550 and 480 g, respectively
The lung CT numbers of drowning cases are mainly influenced by the proportion of inhaled water in the normal lobular lung parenchyma [8]. This means that the airspaces in Type 1 were filled with more water than those of Type 2. Macroscopically, forensic pathologists classify postmortem pulmonary edema as severe, moderate, or mild. The lungs with Type 1 CT findings were usually classified as severe edema. However, the lungs with Type 2 CT findings were classified with a wide range of levels of edema. It is difficult to evaluate the severity of pulmonary edema using macroscopic examination of the lung cut surface in a standard
fashion, because visual assessment is so subjective. Postmortem CT may provide a more objective assessment of pulmonary edema, as the findings are fairly specific. Many forensic textbooks show that one of the characteristic pulmonary findings seen in drowning is called emphysema aquosum [2, 3, 16]. This term is used to describe hyperexpanded and waterlogged lungs with no putrefactive change, implying that the lungs are markedly overinflated, and the findings are the result of prominent expansion by inhaling a combination of air and water while
123
Jpn J Radiol Fig. 4 Postmortem CT image of lungs in drowning classified as Type 1 ? 2 (a). Centrilobular nodules spread out surrounding the airways in the upper lobes (black arrows), and GGO with a superimposed reticular pattern consisting of a combination of interlobular septal thickening and interstitial thickening is widespread in the lower lobes (black arrowhead). The former and latter signs are often observed in Type 2 and Type 1, respectively. This case was macroscopically judged as severe lung edema by the forensic pathologist. Cut sections of the right (b) and left lungs (c) show a bulging cut surface with a brick-red appearance and white foam flowing out similar to Fig. 2b. The right and left lungs weighed 730 and 600 g, respectively
drowning [2, 3, 14, 16]. Previous studies have reported that postmortem ventilation leads to a clearly visible decrease in GGO due to an increase in air in the interstitial space [17, 18]. Thus, a large volume of inhaled/instilled air in the airway, producing a hyperexpanded lung that is the characteristic finding of drowning, should be accompanied by decreased lung CT numbers, and could explain the appearance of faint opacification in Type 2. We believe that emphysema aquosum might be the reason for the Type 2 appearance. The CT appearance of Type 1 is diffuse GGO with thickening of the pulmonary interstitium. The GGO was recognized as two phenomena that often occur simultaneously as observed within the resolution of CT; that is, a mild decrease in the amount of air in the airspaces and a mild increase in fluid content, with preserved visibility of the bronchial and vascular markings because they are outlined by residual alveolar air [8, 19, 20]. Minimal thickening of the alveolar walls and of the parenchymal
interstitium can also cause GGO [8]. In Type 1, the reticular pattern resulting from thickened interlobular septa mainly results from the infiltration of the interstitium with fluid, reflecting pulmonary edema [8, 19, 21]. On the other hand, Type 2 demonstrates faint centrilobular nodules and patchy opacities that may be formed by the fusion of nodules distributed along the airways. The terminal bronchioles are located in the central core of the secondary pulmonary lobules [8, 21]; some alveolar spaces filled with fluid may give this appearance. Therefore, CT number measurements of these two types of drowning lung may be able to distinguish between water in the airways (Type 2) and water in the interstitium (Type 1). Figure 4 shows a case of Type 1 ? 2, meaning that centrilobular nodules surround the airways in both upper lobes and reticular interstitial patterns involve both lower lobes. In this case, according to the police investigation, the body was found floating prone in a river. At autopsy, a large amount of white frothy fluid was found in the airway, with severe
123
Jpn J Radiol
Fig. 5 Postmortem CT image of lungs classified as consolidation Type 3. Irregular consolidation is widespread
Fig. 7 Postmortem CT image of lungs of the unclassified type. The image shows isolated hazy opacities in the right upper lobe (white arrow) and band-like GGO in the left upper lobe (white arrowhead). The other lung parenchyma appears normal
Table 1 Classification table of type of lung findings on postmortem computed tomography (CT) Type of lung findings
Number of cases
CT number ± SD/HU
Type 1
31 (34 %)
-471 ± 85
Type 2
38 (41 %)
-695 ± 66
Type 1 ? 2
10 (11 %)
-606 ± 85
Type 3 (consolidation)
5 (5 %)
-242 ± 162
Type 4 (emphysema/fibrosis)
4 (4 %)
-591 ± 41
Unclassifiable
4 (4 %)
-551 ± 200
Total
92
SD standard deviation, HU Hounsfield unit
Fig. 6 Postmortem CT image of lungs in drowning classified as emphysema and/or fibrosis type. Bullous emphysema and interstitial fibrosis are observed in the lungs. Massive bilateral pleural effusions are also seen
pulmonary edema and a moderate number of diatoms in the lung periphery. Some possible explanations for Type 1 ? 2 may be postmortem change, a transition from Type 2 to Type 1, independent mechanisms in the different lobes, and the intermingling of them. We think further investigation is needed to elucidate the etiology of Type 1 ? 2.
123
Consolidation is defined as an increase in lung attenuation with obscuration of the underlying vessels [8], indicating a decrease in the amount of air in the airspaces and replacement by fluid or other substances. The five cases of consolidation (Type 3) had undergone transfusion during the resuscitation procedure. The autopsy findings of the lungs demonstrated marked edema with little aeration. In this report, we classified it as a special type related to drowning, but to conclude, it may be nonspecific edema due to transfusion therapy. The emphysema and/or fibrosis of Type 4 was histologically confirmed. The lungs did not show the typical findings of drowning, although the death scene was demonstrative. Thus, the CT findings suggested preexisting
Jpn J Radiol
References
Fig. 8 Box plot of the CT number distribution by three Types of CT appearance. The median values as drawn lines in the boxes of each Type (Type 1, Type 1 ? 2, and Type 2) were -492, -624, and -709 HU, respectively. The top and bottom of the boxes indicate the upper and lower quartiles, respectively. Significant differences were observed between each CT pattern Type by the Tukey–Kramer method
pulmonary emphysema and fibrosis, but were unhelpful in diagnosing drowning. When a body is found submerged in water, one of the most important questions is whether the victim inhaled water or not. The autopsy rate in Japan is very low compared with many other developed countries. Many bodies are not autopsied and the cause of death is decided only by external examination. In this article, we report the CT images of drowning cases mainly classified into three main types and two minor types, with a few unclassifiable. This study shows that postmortem CT images can provide significant information about drowning. When one uses postmortem lung CT images to diagnose whether the decedent died by drowning, the remaining issue is the existence of unclassified cases. In this study, unclassified cases amounted to only 4.3 % of total cases. However, when postmortem lung CT shows some image patterns other than the five types, death by drowning cannot be excluded. To clarify why these unique images appear in drowning cases, further comparisons between postmortem lung CT images and autopsy findings are necessary. Conflict of interest disclosed.
No potential conflicts of interest were
1. World Health Organization Fact sheet N347. Drowning. Geneva, Switzerland: WHO; 2012. 2. Spitz DJ. Investigation of bodies in water. In: Spitz WU, editor. Medico-legal investigation of death: guidelines for the application of pathology to crime investigation. 4th ed. Spring-field: Charles C Thomas; 2006. p. 846–81. 3. Saukko P, Knight B. Immersion deaths. In: Saukko P, Knight B, editors. Knight’s forensic pathology. 3rd ed. London: Edward Arnold; 2004. p. 395–411. 4. Christe A, Aghayev E, Jackowski C, Thali MJ, Vock P. Drowning-post-mortem imaging findings by computed tomography. Eur Radiol. 2008;18:283–90. 5. Persson A, Lindblom M, Jackowski C. A state-of-the-art pipeline for postmortem CT and MRI visualization: from data acquisition to interactive image interpretation at autopsy. Acta Radiol. 2011;52:522–36. 6. O’Donnell C, Woodford N. Post-mortem radiology––a new subspeciality? Clin Radiol. 2008;63:1189–94. 7. Kawasumi Y, Kawabata T, Sugai Y, Usui A, Hosokai Y, Sato M, et al. Assessment of the relationship between drowning and fluid accumulation in the paranasal sinuses on post-mortem computed tomography. Eur J Radiol. 2012;81:3953–5. 8. Verschakelen JA, De Wever W. Computed tomography of the lung. A pattern approach. Berlin: Springer; 2007. 9. Levy AD, Harcke HT, Getz JM, et al. Virtual autopsy: two- and three-dimensional multidetector CT findings in drowning with autopsy comparison. Radiology. 2007;243:862–8. 10. Kim KI, Lee KN, Tomiyama N, Johkoh T, Ichikado K, Kim CW, et al. Near drowning: thin-section CT findings in six patients. J Comput Assist Tomogr. 2000;24:562–6. 11. Hunter TB, Whitehouse WM. Fresh-water near-drowning: radiological aspects. Radiology. 1974;112:51–6. 12. Franquet T, Gime´nez A, Roso´n N, Torrubia S, Sabate´ JM, Pe´rez C. Aspiration diseases: findings, pitfalls, and differential diagnosis. Radiographics. 2000;20:673–85. 13. Shiotani S, Kohno M, Ohashi N, Yamazaki K, Nakayama H, Watanabe K, et al. Non-traumatic postmortem computed tomographic (PMCT) findings of the lung. Forensic Sci Int. 2004;139:39–48. 14. DiMaio VJ, DiMaio D. Forensic pathology. 2nd ed. Boca Raton: CRC Press; 2001. 15. Funayama M, Aoki Y, Sebetan IM, Sagisaka K. Detection of diatoms in blood by a combination of membrane filtering and chemical digestion. Forensic Sci Int. 1987;34:175–82. 16. Lunetta P, Modell JH. Macroscopical, microscopical, and laboratory findings in drowning victims: a comprehensive review. In: Tsokos M, editor. Forensic pathology reviews, vol. 3. Totowa: Humana Press; 2005. p. 3–77. 17. Germerott T, Preiss US, Ebert LC, Ruder TD, Ross S, Flach PM, et al. A new approach in virtopsy: postmortem ventilation in multislice computed tomography. Leg Med (Tokyo). 2010;12:276–9. 18. Germerott T, Flach PM, Preiss US, Ross SG, Thali MJ. Postmortem ventilation: a new method for improved detection of pulmonary pathologies in forensic imaging. Leg Med (Tokyo). 2012;14:223–8. 19. Storto ML, Kee ST, Golden JA, Webb WR. Hydrostatic pulmonary edema: high-resolution CT findings. Am J Roentgenol. 1995;165:817–20. 20. Gruden JF, Webb WR, Warnock M. Centrilobular opacities in the lung on high-resolution CT: diagnostic considerations and pathologic correlation. Am J Roentgenol. 1994;162:569–74. 21. Webb WR. Thin-section CT of the secondary pulmonary lobule: anatomy and the image––the 2004 Fleischner lecture. Radiology. 2006;239:322–38.
123
