European Journal of Radiology 84 (2015) 643–651
Contents lists available at ScienceDirect
European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad
Racking the brain: Detection of cerebral edema on postmortem computed tomography compared with forensic autopsy Nicole Berger a,b , Garyfalia Ampanozi a , Wolf Schweitzer a , Steffen G. Ross a , Dominic Gascho a , Thomas D. Ruder a,c , Michael J. Thali a , Patricia M. Flach a,b,∗ a
Institute of Forensic Medicine, Virtopsy, University of Zurich, Winterthurerstrasse 190/52, 8057 Zurich, Switzerland Institute of Diagnostic and Interventional Radiology, University Hospital of Zurich, Raemistrasse 100, 8091 Zurich, Switzerland c Institute of Diagnostic, Interventional and Pediatric Radiology, University Hospital of Bern, Freiburgstrasse, 3010 Bern, Switzerland b
a r t i c l e
i n f o
Article history: Received 20 July 2014 Received in revised form 11 December 2014 Accepted 15 December 2014 Keywords: Forensic radiology Virtopsy Intoxication Postmortem computed tomography (PMCT) Cerebral edema
a b s t r a c t Purpose: The purpose of this study was to compare postmortem computed tomography with forensic autopsy regarding their diagnostic reliability of differentiating between pre-existing cerebral edema and physiological postmortem brain swelling. Materials and methods: The study collective included a total of 109 cases (n = 109/200, 83 male, 26 female, mean age: 53.2 years) and were retrospectively evaluated for the following parameters (as related to the distinct age groups and causes of death): tonsillar herniation, the width of the outer and inner cerebrospinal fluid spaces and the radiodensity measurements (in Hounsfield Units) of the gray and white matter. The results were compared with the findings of subsequent autopsies as the gold standard for diagnosing cerebral edema. p-Values 20 Hounsfield Units), and the gray to white matter ratio was >1.58 when leukoencephalopathy was excluded. Conclusions: Despite normal postmortem changes, generalized brain edema can be differentiated on postmortem computed tomography, and white and gray matter Hounsfield measurements help to determine the cause of death in cases of intoxication or asphyxia. Racking the brain about feasible applications for a precise and reliable brain diagnostic forensic radiology method has just begun. © 2015 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Over the last decade, state-of-the-art forensic investigations have established PMCT as a triage tool, a valuable supplement to autopsy and a replacement for autopsy in selected cases [1–3]. However, forensic radiology is a new field and as such offers narrow diagnostic standards in postmortem applications. There is currently
Abbreviations: CSF, cerebrospinal fluid; DGW, difference of Hounsfield Units between gray and white matter; GM, gray matter; GWR, gray-to-white matter ratio of Hounsfield Units; HU, Hounsfield Units; PMCT, postmortem computed tomography; ROI, region of interest; WM, white matter; SD, standard deviation. ∗ Corresponding author at: Institute of Forensic Medicine, Virtopsy, University of Zurich, Winterthurerstrasse 190/52, 8057 Zurich, Switzerland. Tel.: +41 44 635 5611; fax: +41 44 635 6851. E-mail address: patricia.fl[email protected] (P.M. Flach). http://dx.doi.org/10.1016/j.ejrad.2014.12.014 0720-048X/© 2015 Elsevier Ireland Ltd. All rights reserved.
no established educational curriculum for PMCT and scarce educational material on forensic radiology [4]. In a clinical setting, intracranial edema is associated with diagnostic criteria such as cerebral and cerebellar herniation, ventricular compression, the lack of a distinction between the basilar cisterns and outer CSF spaces, the effacement of sulci and gyri, and a lack of the GM/WM interface on CT [5–12]. Moreover, in living patients the severity of brain edema has been described using a decreasing ratio of attenuation between GM and WM [8,9,11]. The position of the cerebellar tonsils as related to the foramen magnum is delineated by McRae’s line (a line drawn from the basion, which is the inferior tip of the clivus, to the posterior lip of the opisthion). The tonsils are usually located above McRae’s line [13]; however, in rare cases, a lower tonsil position is described as being normal [14]. Differences between brain edema as a cause death or as a contributing factor in a lethal case are difficult to assess using PMCT
644
N. Berger et al. / European Journal of Radiology 84 (2015) 643–651
[2,15–17]. Radiologists tend to use the same criteria for both living and deceased patients in assessing brain edema. However, the differentiation between a pre-existing antemortem (or agonal) brain edema and the typical postmortem generalized brain swelling that occurs immediately after death is challenging when using these clinically known criteria. Typical postmortem brain swelling appears to initially correspond to a cytotoxic edema, which starts immediately at the time of death, aggravates over time and precipitates concomitant decay such as putrefaction and autolysis [12]. The ongoing loss of differentiation between the WM and GM is predominantly attributed to energy failure (deprivation from oxygen and glucose). This disables the sodium–potassium membrane adenosine triphosphatase pump system with consecutive intracellular water accumulation [18]. A subsequent swelling of the cells sets in, resulting in a blurring of the WM and GM on PMCT [2]. In the absence of adenosine triphosphatase, the muscular filaments become permanently complexed, and rigor mortis sets in until decomposition occurs [18]. However, at some point of decay, vasogenic, cytotoxic, osmotic and interstitial brain edema may be present simultaneously. In forensic autopsy, there are criteria for antemortem cerebral edema. These criteria are primarily based on the increased cerebral weight caused by edema, as well as the subjective assessment of flattened gyri and filled sulci as well as swollen hippocampus, herniated cerebellar tonsils and a midline shift in cases in which the edema is unilateral [19]. In this scenario, the normal, sharp demarcation between the GM and WM is lost and the WM appears to be paler and softer in consistency [18]. The literature states that at least 1 h and at best 4 h of survival are needed to detect unequivocal histopathological changes in the brain tissue with a preferable short postmortem interval, in order not to overlay subtle findings by autolysis [19]. The challenge presented by cerebral swelling on PMCT is to differentiate between normal physiological postmortem changes and pre-existing pathology (antemortem or agonal developed brain edema). The purpose of this study was to compare postmortem computed tomography with forensic autopsy regarding their diagnostic reliability to differentiate between pre-existing cerebral edema and physiological postmortem brain swelling and to establish diagnostic criteria for intoxication or asphyxia as the cause of death.
2. Materials and methods 2.1. Study subjects The department of the public prosecutor approved the study. In retrospect, 200 consecutive subjects with pre-autopsy whole-body PMCT scans, separate head scans and full forensic autopsy, were initially included in the study (from July 2011 to April 2013). Of the 200 cases, 91 cases had to be excluded because of intracranial hemorrhage (n = 20), cerebral primary tumor or metastasis (n = 2), focal/unilateral edema (n = 5), age under 18 years (n = 6) (excluded due to the absence of standard normal brain weight tables in this age), extensive traumatic brain injury (n = 19), thermal impact (charred, hypothermic or frozen bodies) (n = 5), postmortem decomposition (n = 24), hydrocephalus (n = 7) or technical artifacts (n = 3), leaving 109 cases (83 male, 26 female) for analysis. The mean age of the included subjects was 53.2 years (range 20–88 years). The mean time, which elapsed between time of death and PMCT scan (death-to-PMCT interval), was 31 h and varied from 4 h to 199 h. The causes of death were central regulatory failure (n = 4), ex/ensanguination (n = 12), cardiac failure (n = 48), asphyxia (n = 9), multi-organ/system failure (n = 9), intoxication (n = 24), hypothermia (n = 1), tumor/metastasis (n = 1) and metabolic (n = 1).
2.2. Groups The subjects were grouped by age for analysis according to the age-dependent changes of the brain parenchyma and the CSF spaces: group 1 (18–35 years, n = 18, 17.3%), group 2 (36–50 years, n = 30, 27.2%), group 3 (51–70 years, n = 42, 38.2%) and group 4 (>70 years, n = 19, 17.3%). As subgroups, the cases of intoxication (n = 24) and intoxication and/or suffocation (n = 35) were separated from other causes of death, whereas cases with leukoencephalopathy (n = 21) were excluded.
2.3. Forensic autopsy The autopsy included the external inspection and opening of all three body cavities (skull, thorax and abdomen). The brain weight (g) was routinely measured and documented in each case during autopsy and was supervised by at least one board-certified forensic pathologist. As a correlate of brain edema, the relative deviation of the brain weight over the age-dependent normal values [20] as well as the reduced tissue consistency and flattening of the gyri [21] were used. The autopsy findings served as the reference standard. Histology was not performed for the diagnosis of brain edema. Toxicology was performed in cases with suspected drug abuse or intoxication. The time elapsed between time of death and autopsy (death-toautopsy interval) was 41 h and ranged from 5 h to 201 h.
2.4. CT data acquisition and image reconstruction Image acquisition was performed on a dual-source CT scanner (SOMATOM Flash Definition, Siemens, Forchheim, Germany). The scans were performed after the arrival of the deceased at the Institute of Forensic Medicine and prior to autopsy with 120 kV and automatic dose modulation (CARE Dose4DTM , Siemens, Forchheim, Germany). The imaging included a complete whole-body scan (from head to toe with lowered arms) with an extended field of view (slice thickness of 2 mm), a separate head and neck scan (slice thickness 0.6 mm, increment 0.4 mm, reference mAs 800) with an adjusted field of view (FoV, maximum 300 mm), a separate thorax/abdomen scan with elevated arms (slice thickness 1 mm) with soft tissue and bone window/lung window with (respectively) soft and hard kernel reconstructions. In detail, the image reconstruction of the head and neck further included orbitomeatal and symmetrically aligned axial thick-sliced images (4 mm, 3 mm increment) in a cerebral window in a soft kernel via a 3-dimensional reconstruction tool for multiplanar reconstruction (MPR). Additional multiplanar and 3-dimensional reconstructions were performed at a multimodality workplace (LEONARDO, SynGo, Siemens Medical Solutions, Erlangen, Germany).
2.5. PMCT data analysis Retrospective image evaluation was performed on a Picture and Archiving Communications System (IDS 7, Sectra, Linköping, Sweden). A board-certified radiologist with 6 years of experience in postmortem forensic radiology reviewed the images and was blinded to the autopsy results and case circumstances. The radiological data analysis regarding the absence or presence of brain edema included the following parameters: 1. Measurements of the inner CSF spaces – the left and right side ventricle.
N. Berger et al. / European Journal of Radiology 84 (2015) 643–651
The inner CSF space was graded by measuring (in mm) the width of the right and the left lateral ventricles at the level of the largest extent in an axial plane on the thick slices (4 mm). 2. Measurements of the inner CSF spaces – the temporal horn. The temporal horn on both sides was graded based on a 3-point Likert scale (1 = not delimitable, 2 = delimitable, 3 = expanded) on the axial image stack (4 mm) (Fig. 1). 3. Measurements of the outer CSF spaces. The outer CSF space was evaluated by estimating the width based on a 3-point Likert scale (1 = narrow, 2 = normal, 3 = expanded) in an axial plane just above the lateral ventricles (Fig. 1). 4. Measurement of the HU of the GM and WM. Two regions of interest were defined on axial imaging: (1) the WM at the centrum semiovale level (defined as the Image 5 mm above the lateral ventricles) and (2) GM at the cortex of the frontal lobe level (defined as the Image 5 mm below the centrum semiovale level). The measuring cursor was configured for a ROI as a 28-mm2 (radius 3 mm) circular surface, and the slice thickness was 4 mm. The HU values were taken from the medial cortex to avoid alterations induced by beam-hardening bone artifacts. The average of the measurements (three measurements for each site) was recorded as the HU value for the GM and WM. The GM/WM ratio (GWR = GM/WM) and difference in HU values between the GM and WM (DGW = GM − WM) were calculated for all cases and were calculated separately for intoxication and intoxication/asphyxia. To reduce bias, cases
645
with periventricular leukoencephalopathy were excluded in a subgroup evaluation (Fig. 2). 5. Measurements of the cerebellar tonsils. In the multiplanar reconstruction, a line according to McRae’s line was drawn on the sagittal plane in the bone window using a caliper tool. A thin-slice image reconstruction was used for reformations (0.6 mm) and evaluated in a cerebrum window in the soft kernel and a thick sliced multiplanar reconstruction (4 mm) to reduce noise in the posterior fossa and brain stem region. The distance (mm) from the most basal tip of the tonsils on each side to the drawn McRae’s line was measured (Fig. 3). 6. Visual assessment of brain edema. The rater assessed if there were PMCT signs of pre-existing brain edema. 7. Visual assessment of intoxication. After the first complete data analysis, the rater detected a difference in the attenuation ratio between the GM and WM (GWR) in some cases; in this circumstance, the rater again assessed all scans for visual indicators of intoxication and/or asphyxia as the cause of death. The visual indicator was adistinct hypodense depiction of the WM compared to the GM (concomitant with brain edema and the visually increased GWR and DGW). Various methods have been evaluated in the literature including subtracting the values of GM and WM from the CSF or measuring both against a skull water phantom or experimental investigations with rats and gelatin gel models [22,23].
Fig. 1. Upper row: grading of the inner cerebrospinal fluid spaces at the temporal horns. (A) Expanded temporal horns, no cerebral edema with normal postmortem changes. (B) Delimitable temporal horns: no cerebral edema with normal postmortem changes. (C) Not delimitable temporal horns corresponding to agonal or antemortal cerebral edema. Lower row: grading of the outer CSF spaces. (D) Outer CSF slightly expanded on PMCT. (E) Normal postmortem outer CSF spaces. (F) Narrow outer CSF spaces.
646
N. Berger et al. / European Journal of Radiology 84 (2015) 643–651
Fig. 2. Measurement of the attenuation of the white and gray matter (A: no edema, B: edema without intoxication, C: edema with intoxication). (A) A 36-year-old male who died from ex- and ensanguination and fat embolism in a case of suicide. The GM (gray matter) measures 47 HU (Hounsfield Units); the WM (white matter) measures 34.4 HU, resulting in a GWR (gray matter to white matter ratio) of 1.37 and a DGW (difference between the gray matter and white matter) of 12.6 HU. This case showed no brain edema despite typical postmortem changes. (B) A 38-year-old female who succumbed to a natural cause of death. She suffered from an inguinal squamous cell carcinoma with pulmonary embolism and therefore cardiac arrest. The GM measured 49.6 HU, the WM measured 36.6 HU, the GWR was 1.35 and the DGW was 13 HU. These measurements did not enable the differentiation from normal postmortem changes in the brain based on this study’s HU measurements. (C) A 40-year-old female who died from central regulatory failure due to intoxication with cocaine and methadone. Note the distinct hypodense appearance of the WM (28.6 HU) compared with the GM (50.8 HU) in this distinct brain edema. The GWR of 1.77 and the DGW 22.2 HU supported the visual determination of intoxication or asphyxia.
Fig. 3. (A) McRae’s line (white line) extends from the basion to the opisthion in the sagittal bone window and defines the lines of the downward herniation of the tonsils. (B) Tonsillar herniation is then measured perpendicular to McRae’s line (in this case 1 cm below McRae’s line) on a thick-sliced sagittal cerebrum window. (C) Corresponding axial slice of the bilateral herniated tonsils in this 50-year-old male with asphyxia and central regulatory failure presenting with brain edema. Manner of death was suicide. (D) Macroscopic photograph of the tonsillar impressions due to elevated intracranial pressure.
N. Berger et al. / European Journal of Radiology 84 (2015) 643–651
647
Fig. 4. (A) Regression line for GWR and death-to-PMCT interval with statistical significance (p value
Contents lists available at ScienceDirect
European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad
Racking the brain: Detection of cerebral edema on postmortem computed tomography compared with forensic autopsy Nicole Berger a,b , Garyfalia Ampanozi a , Wolf Schweitzer a , Steffen G. Ross a , Dominic Gascho a , Thomas D. Ruder a,c , Michael J. Thali a , Patricia M. Flach a,b,∗ a
Institute of Forensic Medicine, Virtopsy, University of Zurich, Winterthurerstrasse 190/52, 8057 Zurich, Switzerland Institute of Diagnostic and Interventional Radiology, University Hospital of Zurich, Raemistrasse 100, 8091 Zurich, Switzerland c Institute of Diagnostic, Interventional and Pediatric Radiology, University Hospital of Bern, Freiburgstrasse, 3010 Bern, Switzerland b
a r t i c l e
i n f o
Article history: Received 20 July 2014 Received in revised form 11 December 2014 Accepted 15 December 2014 Keywords: Forensic radiology Virtopsy Intoxication Postmortem computed tomography (PMCT) Cerebral edema
a b s t r a c t Purpose: The purpose of this study was to compare postmortem computed tomography with forensic autopsy regarding their diagnostic reliability of differentiating between pre-existing cerebral edema and physiological postmortem brain swelling. Materials and methods: The study collective included a total of 109 cases (n = 109/200, 83 male, 26 female, mean age: 53.2 years) and were retrospectively evaluated for the following parameters (as related to the distinct age groups and causes of death): tonsillar herniation, the width of the outer and inner cerebrospinal fluid spaces and the radiodensity measurements (in Hounsfield Units) of the gray and white matter. The results were compared with the findings of subsequent autopsies as the gold standard for diagnosing cerebral edema. p-Values 20 Hounsfield Units), and the gray to white matter ratio was >1.58 when leukoencephalopathy was excluded. Conclusions: Despite normal postmortem changes, generalized brain edema can be differentiated on postmortem computed tomography, and white and gray matter Hounsfield measurements help to determine the cause of death in cases of intoxication or asphyxia. Racking the brain about feasible applications for a precise and reliable brain diagnostic forensic radiology method has just begun. © 2015 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Over the last decade, state-of-the-art forensic investigations have established PMCT as a triage tool, a valuable supplement to autopsy and a replacement for autopsy in selected cases [1–3]. However, forensic radiology is a new field and as such offers narrow diagnostic standards in postmortem applications. There is currently
Abbreviations: CSF, cerebrospinal fluid; DGW, difference of Hounsfield Units between gray and white matter; GM, gray matter; GWR, gray-to-white matter ratio of Hounsfield Units; HU, Hounsfield Units; PMCT, postmortem computed tomography; ROI, region of interest; WM, white matter; SD, standard deviation. ∗ Corresponding author at: Institute of Forensic Medicine, Virtopsy, University of Zurich, Winterthurerstrasse 190/52, 8057 Zurich, Switzerland. Tel.: +41 44 635 5611; fax: +41 44 635 6851. E-mail address: patricia.fl[email protected] (P.M. Flach). http://dx.doi.org/10.1016/j.ejrad.2014.12.014 0720-048X/© 2015 Elsevier Ireland Ltd. All rights reserved.
no established educational curriculum for PMCT and scarce educational material on forensic radiology [4]. In a clinical setting, intracranial edema is associated with diagnostic criteria such as cerebral and cerebellar herniation, ventricular compression, the lack of a distinction between the basilar cisterns and outer CSF spaces, the effacement of sulci and gyri, and a lack of the GM/WM interface on CT [5–12]. Moreover, in living patients the severity of brain edema has been described using a decreasing ratio of attenuation between GM and WM [8,9,11]. The position of the cerebellar tonsils as related to the foramen magnum is delineated by McRae’s line (a line drawn from the basion, which is the inferior tip of the clivus, to the posterior lip of the opisthion). The tonsils are usually located above McRae’s line [13]; however, in rare cases, a lower tonsil position is described as being normal [14]. Differences between brain edema as a cause death or as a contributing factor in a lethal case are difficult to assess using PMCT
644
N. Berger et al. / European Journal of Radiology 84 (2015) 643–651
[2,15–17]. Radiologists tend to use the same criteria for both living and deceased patients in assessing brain edema. However, the differentiation between a pre-existing antemortem (or agonal) brain edema and the typical postmortem generalized brain swelling that occurs immediately after death is challenging when using these clinically known criteria. Typical postmortem brain swelling appears to initially correspond to a cytotoxic edema, which starts immediately at the time of death, aggravates over time and precipitates concomitant decay such as putrefaction and autolysis [12]. The ongoing loss of differentiation between the WM and GM is predominantly attributed to energy failure (deprivation from oxygen and glucose). This disables the sodium–potassium membrane adenosine triphosphatase pump system with consecutive intracellular water accumulation [18]. A subsequent swelling of the cells sets in, resulting in a blurring of the WM and GM on PMCT [2]. In the absence of adenosine triphosphatase, the muscular filaments become permanently complexed, and rigor mortis sets in until decomposition occurs [18]. However, at some point of decay, vasogenic, cytotoxic, osmotic and interstitial brain edema may be present simultaneously. In forensic autopsy, there are criteria for antemortem cerebral edema. These criteria are primarily based on the increased cerebral weight caused by edema, as well as the subjective assessment of flattened gyri and filled sulci as well as swollen hippocampus, herniated cerebellar tonsils and a midline shift in cases in which the edema is unilateral [19]. In this scenario, the normal, sharp demarcation between the GM and WM is lost and the WM appears to be paler and softer in consistency [18]. The literature states that at least 1 h and at best 4 h of survival are needed to detect unequivocal histopathological changes in the brain tissue with a preferable short postmortem interval, in order not to overlay subtle findings by autolysis [19]. The challenge presented by cerebral swelling on PMCT is to differentiate between normal physiological postmortem changes and pre-existing pathology (antemortem or agonal developed brain edema). The purpose of this study was to compare postmortem computed tomography with forensic autopsy regarding their diagnostic reliability to differentiate between pre-existing cerebral edema and physiological postmortem brain swelling and to establish diagnostic criteria for intoxication or asphyxia as the cause of death.
2. Materials and methods 2.1. Study subjects The department of the public prosecutor approved the study. In retrospect, 200 consecutive subjects with pre-autopsy whole-body PMCT scans, separate head scans and full forensic autopsy, were initially included in the study (from July 2011 to April 2013). Of the 200 cases, 91 cases had to be excluded because of intracranial hemorrhage (n = 20), cerebral primary tumor or metastasis (n = 2), focal/unilateral edema (n = 5), age under 18 years (n = 6) (excluded due to the absence of standard normal brain weight tables in this age), extensive traumatic brain injury (n = 19), thermal impact (charred, hypothermic or frozen bodies) (n = 5), postmortem decomposition (n = 24), hydrocephalus (n = 7) or technical artifacts (n = 3), leaving 109 cases (83 male, 26 female) for analysis. The mean age of the included subjects was 53.2 years (range 20–88 years). The mean time, which elapsed between time of death and PMCT scan (death-to-PMCT interval), was 31 h and varied from 4 h to 199 h. The causes of death were central regulatory failure (n = 4), ex/ensanguination (n = 12), cardiac failure (n = 48), asphyxia (n = 9), multi-organ/system failure (n = 9), intoxication (n = 24), hypothermia (n = 1), tumor/metastasis (n = 1) and metabolic (n = 1).
2.2. Groups The subjects were grouped by age for analysis according to the age-dependent changes of the brain parenchyma and the CSF spaces: group 1 (18–35 years, n = 18, 17.3%), group 2 (36–50 years, n = 30, 27.2%), group 3 (51–70 years, n = 42, 38.2%) and group 4 (>70 years, n = 19, 17.3%). As subgroups, the cases of intoxication (n = 24) and intoxication and/or suffocation (n = 35) were separated from other causes of death, whereas cases with leukoencephalopathy (n = 21) were excluded.
2.3. Forensic autopsy The autopsy included the external inspection and opening of all three body cavities (skull, thorax and abdomen). The brain weight (g) was routinely measured and documented in each case during autopsy and was supervised by at least one board-certified forensic pathologist. As a correlate of brain edema, the relative deviation of the brain weight over the age-dependent normal values [20] as well as the reduced tissue consistency and flattening of the gyri [21] were used. The autopsy findings served as the reference standard. Histology was not performed for the diagnosis of brain edema. Toxicology was performed in cases with suspected drug abuse or intoxication. The time elapsed between time of death and autopsy (death-toautopsy interval) was 41 h and ranged from 5 h to 201 h.
2.4. CT data acquisition and image reconstruction Image acquisition was performed on a dual-source CT scanner (SOMATOM Flash Definition, Siemens, Forchheim, Germany). The scans were performed after the arrival of the deceased at the Institute of Forensic Medicine and prior to autopsy with 120 kV and automatic dose modulation (CARE Dose4DTM , Siemens, Forchheim, Germany). The imaging included a complete whole-body scan (from head to toe with lowered arms) with an extended field of view (slice thickness of 2 mm), a separate head and neck scan (slice thickness 0.6 mm, increment 0.4 mm, reference mAs 800) with an adjusted field of view (FoV, maximum 300 mm), a separate thorax/abdomen scan with elevated arms (slice thickness 1 mm) with soft tissue and bone window/lung window with (respectively) soft and hard kernel reconstructions. In detail, the image reconstruction of the head and neck further included orbitomeatal and symmetrically aligned axial thick-sliced images (4 mm, 3 mm increment) in a cerebral window in a soft kernel via a 3-dimensional reconstruction tool for multiplanar reconstruction (MPR). Additional multiplanar and 3-dimensional reconstructions were performed at a multimodality workplace (LEONARDO, SynGo, Siemens Medical Solutions, Erlangen, Germany).
2.5. PMCT data analysis Retrospective image evaluation was performed on a Picture and Archiving Communications System (IDS 7, Sectra, Linköping, Sweden). A board-certified radiologist with 6 years of experience in postmortem forensic radiology reviewed the images and was blinded to the autopsy results and case circumstances. The radiological data analysis regarding the absence or presence of brain edema included the following parameters: 1. Measurements of the inner CSF spaces – the left and right side ventricle.
N. Berger et al. / European Journal of Radiology 84 (2015) 643–651
The inner CSF space was graded by measuring (in mm) the width of the right and the left lateral ventricles at the level of the largest extent in an axial plane on the thick slices (4 mm). 2. Measurements of the inner CSF spaces – the temporal horn. The temporal horn on both sides was graded based on a 3-point Likert scale (1 = not delimitable, 2 = delimitable, 3 = expanded) on the axial image stack (4 mm) (Fig. 1). 3. Measurements of the outer CSF spaces. The outer CSF space was evaluated by estimating the width based on a 3-point Likert scale (1 = narrow, 2 = normal, 3 = expanded) in an axial plane just above the lateral ventricles (Fig. 1). 4. Measurement of the HU of the GM and WM. Two regions of interest were defined on axial imaging: (1) the WM at the centrum semiovale level (defined as the Image 5 mm above the lateral ventricles) and (2) GM at the cortex of the frontal lobe level (defined as the Image 5 mm below the centrum semiovale level). The measuring cursor was configured for a ROI as a 28-mm2 (radius 3 mm) circular surface, and the slice thickness was 4 mm. The HU values were taken from the medial cortex to avoid alterations induced by beam-hardening bone artifacts. The average of the measurements (three measurements for each site) was recorded as the HU value for the GM and WM. The GM/WM ratio (GWR = GM/WM) and difference in HU values between the GM and WM (DGW = GM − WM) were calculated for all cases and were calculated separately for intoxication and intoxication/asphyxia. To reduce bias, cases
645
with periventricular leukoencephalopathy were excluded in a subgroup evaluation (Fig. 2). 5. Measurements of the cerebellar tonsils. In the multiplanar reconstruction, a line according to McRae’s line was drawn on the sagittal plane in the bone window using a caliper tool. A thin-slice image reconstruction was used for reformations (0.6 mm) and evaluated in a cerebrum window in the soft kernel and a thick sliced multiplanar reconstruction (4 mm) to reduce noise in the posterior fossa and brain stem region. The distance (mm) from the most basal tip of the tonsils on each side to the drawn McRae’s line was measured (Fig. 3). 6. Visual assessment of brain edema. The rater assessed if there were PMCT signs of pre-existing brain edema. 7. Visual assessment of intoxication. After the first complete data analysis, the rater detected a difference in the attenuation ratio between the GM and WM (GWR) in some cases; in this circumstance, the rater again assessed all scans for visual indicators of intoxication and/or asphyxia as the cause of death. The visual indicator was adistinct hypodense depiction of the WM compared to the GM (concomitant with brain edema and the visually increased GWR and DGW). Various methods have been evaluated in the literature including subtracting the values of GM and WM from the CSF or measuring both against a skull water phantom or experimental investigations with rats and gelatin gel models [22,23].
Fig. 1. Upper row: grading of the inner cerebrospinal fluid spaces at the temporal horns. (A) Expanded temporal horns, no cerebral edema with normal postmortem changes. (B) Delimitable temporal horns: no cerebral edema with normal postmortem changes. (C) Not delimitable temporal horns corresponding to agonal or antemortal cerebral edema. Lower row: grading of the outer CSF spaces. (D) Outer CSF slightly expanded on PMCT. (E) Normal postmortem outer CSF spaces. (F) Narrow outer CSF spaces.
646
N. Berger et al. / European Journal of Radiology 84 (2015) 643–651
Fig. 2. Measurement of the attenuation of the white and gray matter (A: no edema, B: edema without intoxication, C: edema with intoxication). (A) A 36-year-old male who died from ex- and ensanguination and fat embolism in a case of suicide. The GM (gray matter) measures 47 HU (Hounsfield Units); the WM (white matter) measures 34.4 HU, resulting in a GWR (gray matter to white matter ratio) of 1.37 and a DGW (difference between the gray matter and white matter) of 12.6 HU. This case showed no brain edema despite typical postmortem changes. (B) A 38-year-old female who succumbed to a natural cause of death. She suffered from an inguinal squamous cell carcinoma with pulmonary embolism and therefore cardiac arrest. The GM measured 49.6 HU, the WM measured 36.6 HU, the GWR was 1.35 and the DGW was 13 HU. These measurements did not enable the differentiation from normal postmortem changes in the brain based on this study’s HU measurements. (C) A 40-year-old female who died from central regulatory failure due to intoxication with cocaine and methadone. Note the distinct hypodense appearance of the WM (28.6 HU) compared with the GM (50.8 HU) in this distinct brain edema. The GWR of 1.77 and the DGW 22.2 HU supported the visual determination of intoxication or asphyxia.
Fig. 3. (A) McRae’s line (white line) extends from the basion to the opisthion in the sagittal bone window and defines the lines of the downward herniation of the tonsils. (B) Tonsillar herniation is then measured perpendicular to McRae’s line (in this case 1 cm below McRae’s line) on a thick-sliced sagittal cerebrum window. (C) Corresponding axial slice of the bilateral herniated tonsils in this 50-year-old male with asphyxia and central regulatory failure presenting with brain edema. Manner of death was suicide. (D) Macroscopic photograph of the tonsillar impressions due to elevated intracranial pressure.
N. Berger et al. / European Journal of Radiology 84 (2015) 643–651
647
Fig. 4. (A) Regression line for GWR and death-to-PMCT interval with statistical significance (p value
