Autopsy Imaging Research Articles

Inconsistency between Human Observation and Deep Learning Models: Assessing Validity of Postmortem Computed Tomography Diagnosis of Drowning

Drowning diagnosis is a complicated process in the autopsy, even with the assistance of autopsy imaging and the on-site information from where the body was found. Previous studies have developed well-performed deep learning (DL) models for drowning diagnosis. However, the validity of the DL models was not assessed, raising doubts about whether the learned features accurately represented the medical findings observed by human experts. In this paper, we assessed the medical validity of DL models that had achieved high classification performance for drowning diagnosis. This retrospective study included autopsy cases aged 8–91 years who underwent postmortem computed tomography between 2012 and 2021 (153 drowning and 160 non-drowning cases). We first trained three deep learning models from a previous work and generated saliency maps that highlight important features in the input. To assess the validity of models, pixel-level annotations were created by four radiological technologists and further quantitatively compared with the saliency maps. All the three models demonstrated high classification performance with areas under the receiver operating characteristic curves of 0.94, 0.97, and 0.98, respectively. On the other hand, the assessment results revealed unexpected inconsistency between annotations and models’ saliency maps. In fact, each model had, respectively, around 30%, 40%, and 80% of irrelevant areas in the saliency maps, suggesting the predictions of the DL models might be unreliable. The result alerts us in the careful assessment of DL tools, even those with high classification performance.

Optimization of the fluid-attenuated inversion recovery (FLAIR) imaging for use in autopsy imaging of the brain region using synthetic MRI.

The failure of cerebrospinal fluid (CSF) signal suppression in postmortem fluid-attenuated inversion recovery (FLAIR) of the brain is a problem. The present study was to clarify the relationship between the temperature of deceased persons and CSF T1, and to optimize the postmortem brain FLAIR imaging method using synthetic MRI. Forehead temperature was measured in 15 deceased persons. Next, synthetic MRI of the brain was performed, the CSF T1 was measured, and the optimal TI was calculated. Two types of FLAIR images were obtained with the clinical and optimal TI. The relationship between forehead temperature and the CSF T1 and optimal TI was evaluated. The optimized FLAIR images were physically and visually evaluated. The CSF T1 and optimal TI were strongly correlated with forehead temperature. Comparing the average SNR and CNR ratios and visual evaluation scores of the two FLAIR images, those captured with the optimal TI showed statistically lower SNR, higher CNR, and higher visual evaluation scores (p< 0.01). Synthetic MRI enables the quantification of the CSF T1 resulting from postmortem temperature decreases and calculation of the optimal TI, which could aid in improving the failure of CSF signal suppression and in optimizing postmortem brain FLAIR imaging.

A Clue Supporting a Diagnosis of Atlanto-Occipital Dislocation Based on a Traumatic Vacuum Phenomenon.

Sir, The high spatial and densitometric resolution of computed tomography (CT) means that it frequently shows gas (vacuum phenomenon [VP]) in tissues. VP is observed at locations that experience traumatic impact; thus, an analysis of VP may be useful for elucidating the mechanism of injury.[1] The mechanism responsible for the formation of VP is as follows: if an enclosed tissue space is allowed to expand as a rebound phenomenon after an external impact, the volume within the enclosed space will increase. In the setting of expanding volume, the pressure within the space will decrease. The solubility of the gas in the enclosed space will decrease as the pressure of the space decreases. Decreased solubility allows a gas to leave a solution. We herein report a patient in whom VP was detected near the foramen magnum after blunt trauma that resulted in a successful diagnosis of atlanto-occipital dislocation. The patient was a 53-year-old man who suddenly crashed while motorcycling on a passing lane of a national route. He was thrown from the vehicle and hit a guardrail on the median. At 16 min from the first call, emergency medical technicians reached him. He was in cardiac arrest at the time, and the initial rhythm was asystole. Bystander basic life support was absent. He was bleeding from the nose and mouth. He was transported to the grounds of a nearby elementary school and evacuated by a physician-staffed helicopter. On arrival, he remained in cardiac arrest. He underwent resuscitative thoracotomy, but it failed to obtain the return of spontaneous circulation. Autopsy CT showed subarachnoid hemorrhaging at the basal cistern and posterior fossa. A sagittal view of the skull and cervical bone on CT did not show any fracture or dislocation. However, he had several small gas densities at the deep neck area near the posterior foramen magnum [Figure 1], so the complication of atlanto-occipital dislocation was suspected. Autopsy imaging of magnetic resonance imaging showed atlanto-occipital dislocation compressing the junction between the medulla and spinal cord [Figure 1].Figure 1: Sagittal view of computed tomography (CT, left and middle) with bony field and magnetic resonance imaging (MRI, short-tau inversion recovery image) on arrival. The CT showed several small gas densities (white triangle) at the deep neck area near the posterior foramen magnum without fracture or dislocation. MRI showed atlanto-occipital dislocation (black triangle) compressing the junction between the medulla and spinal cord. The airway was shifted frontally by a hematoma (star). CT: Computed tomography, MRI: Magnetic resonance imagingThis is the first case report, wherein VP near the foramen magnum was a clue supporting a diagnosis of atlanto-occipital dislocation. The diagnosis of atlanto-occipital dislocation by imaging is not easy, and indeed, one report showed no correct diagnoses of atlanto-occipital dislocation when patients were in cardiac arrest.[2,3] Overlooking atlanto-occipital dislocation can result in a fatal outcome.[4] Accordingly, physician should strongly suspect the complication of atlanto-occipital dislocation when patients have traumatic VP near the foramen magnum on CT. Declaration of patient consent The authors certify that they have obtained all appropriate patient consent form. In the form, the patient’s family has given his consent for his image and other clinical information to be reported in the journal. The patient’s family understands that his name and initial will not be published and due efforts will be made to conceal his identity, but anonymity cannot be guaranteed. Research quality and ethics statement The authors followed applicable EQUATOR Network (https:// www.equator-network.org/) guidelines, notably the CARE guideline, during the conduct of this report. Financial support and sponsorship This work was supported in part by a Grant-in-Aid for Special Research in Subsidies for ordinary expenses of private schools from The Promotion and Mutual Aid Corporation for Private Schools of Japan. Conflicts of interest There are no conflicts of interest.

Role of autopsy imaging in veterinary forensic medicine: experiences in 39 cases.

While numerous scientific studies have suggested the usefulness of autopsy imaging (Ai) in the field of human forensic medicine, the use of imaging modalities for the purpose of veterinary forensics is currently scant. The current study performed Ai on suspicious dead animals requested by the police department to determine their cause of death. Radiography and/or computed tomography and/or magnetic resonance imaging were performed on 39 suspicious dead animals before necropsy. After diagnostic imaging, pathological examination was performed, with drug testing added as needed. Among the 39 cases, 28, 6, 3, 1, and 1 involved cats, dogs, rabbits, a ferret, and a pigeon, respectively. The major Ai findings included skull and rib fractures, subcutaneous emphysema, pneumothorax, pneumoperitoneum, diaphragmatic hernia, and abdominal rupture. The leading causes of death, determined comprehensively via Ai and pathological reports and drug test results, included traumatic impact, blood loss, poisoning, suffocation, tension pneumothorax, starvation, and drowning, all of which have been strongly suspected to indicate animal abuse by humans. All eight cases of skull fractures and all five cases of poisoning, including suspected poisoning, were of cats. As the numbers of dogs and cats in Japan are almost equal, violence against cats may occur more frequently than dogs. Ai can be a valuable examination tool for veterinary forensic cases. As computed tomography is useful for ruling out fractures that screening computed tomography before necropsy is a more practical option for veterinary forensics.

Research Progress of Forensic Diagnosis Approaches of Early Acute Myocardial Infarction.

The postmortem diagnosis of acute myocardial infarction (AMI), especially the postmortem diagnosis of early AMI that died immediately after onset or within 1 hour, has always been a difficulty in forensic identification. This article reviews the forensic application of diagnosis and analysis methods for AMI postmortem diagnosis including autopsy imaging, histomorphology, immunohisto-chemistry, biochemical marker and molecular biology diagnosis, and explores the feasible scheme of early postmortem diagnosis in AMI.

Role of Autopsy Imaging in Veterinary Forensics: Experiences in 39 Cases

Role of Autopsy Imaging in Veterinary Forensics: Experiences in 39 Cases

Significance of medical intervention for non-traumatic hemorrhagic cardiac tamponade.

The outcomes of patients with cardiac arrest induced by non-traumatic hemorrhagic cardiac tamponade are poor. We retrospectively investigated the significance of medical intervention with pericardiocentesis and/or pericardiotomy for non-traumatic hemorrhagic cardiac tamponade. From January 2013 to April 2021, we retrospectively reviewed the medical charts of all patients with cardiac arrest in a prehospital setting or emergency room due to cardiac tamponade confirmed by an ultrasound examination with or without an invasive procedure (pericardiocentesis and/or pericardiotomy) and computed tomography findings, including those obtained at autopsy imaging. The subjects were divided into two groups: the Intervention (+) group, which included subjects who underwent pericardiocentesis or pericardiotomy, and the Intervention (-) group, which included subjects who did not undergo pericardiocentesis or pericardiotomy. Variables were then compared between the two groups. There were 68 patients with non-traumatic cardiac tamponade. All three survival cases had witnessed collapse, and the initial rhythm was pulseless electrical activity (PEA).There were no statistically significant differences in the sex, age, means of transportation, bystander chest compression, electric shock, or adrenalineor FDP levels between the two groups.However, the number with witnessed collapse, PEA, rupture of the heart;the ratio of obtaining return of spontaneous circulation; and the survival ratio were significantly greater in the Intervention (+) group than in the Intervention (-) group. Based on the results of preliminary study, we hypothesized that invasive medical intervention for patients with cardiac arrest induced by non-traumatic hemorrhagic cardiac tamponade might be useful for obtaining return of spontaneous circulation and a survival outcome, especially for patients with witnessed collapse with PEA as the initial rhythm.

Population-based incidence and outcomes of acute aortic dissection in Japan

Abstract Background The population-based incidence and outcomes of acute aortic dissection (AAD) are still unknown because some patients are already dead on arrival at the hospital, and the accurate diagnosis of AAD is difficult due to the low autopsy rate for patients with cardiopulmonary arrest outside of the hospital. We performed a population-based review of all patients with AAD in a well-defined geographical area in the southern part of Japan between 2016 and 2018. Methods Data of all patients with AAD at our Hospital, which performs medical care for 120,000 residents, were collected retrospectively. The emergency medical service is dedicated to the transfer of all patients in this area to the MPNH. For all patients who were dead on arrival, the diagnosis of AAD was made by autopsy imaging (AI) using computed tomography. The age-adjusted incidence and mortality per 100,000 population were calculated using the Japanese population distribution model in 2015. Results The total incidence of AAD was 79 (type A: 64.5%, n=51). Of those, 60.8% (31/51) of patients with type A and 21.4% (6/28) with type B were dead on arrival and diagnosed by AI. The 30-day mortality rates after the onset of AAD were 74.5% (38/51) in type A and 25.0% (7/28) in type B. Excluding the dead-on-arrival patients, the 30-day mortality rates were 35.0% (7/20) in type A and 4.5% (1/22) in type B. The age-adjusted incidence and mortality of AAD per 100,000 inhabitants were 17.3 (type A: 11.3, type B: 6.0) and 9.6 (type A: 8.4, type B: 1.2), respectively. Both values were significantly higher in men than in women. Conclusions The population-based survey of emergency medical care for AAD showed that the age-adjusted incidence of AAD was two-fold higher than in previous reports, and the actual mortality rates after AAD onset were markedly higher due to the high incidence of dead-on-arrival. Funding Acknowledgement Type of funding sources: None. Representative CT images of type A AAD

Population-based incidence and outcomes of acute aortic dissection in Japan.

The population-based incidence and outcomes of acute aortic dissection (AAD) are still unknown because some patients are already dead on arrival, and the accurate diagnosis of AAD is difficult due to the low autopsy rate. We performed a population-based review of all patients with AAD in a well-defined geographical area in Japan between 2016 and 2018. Data of all patients with AAD at Miyazaki Prefectural Nobeoka Hospital (MPNH), which performs medical care for 120 000 residents, were collected retrospectively. The emergency medical service is dedicated to the transfer of all patients in this area to the MPNH. For all patients who were dead on arrival, the diagnosis of AAD was made by autopsy imaging (AI) using computed tomography. The age-adjusted incidence and mortality per 100 000 population were calculated using the Japanese population distribution model in 2015. The total incidence of AAD was 79 (type A: 64.5%, n = 51). Of those, 60.8% (31/51) of patients with type A and 21.4% (6/28) with type B were dead on arrival and diagnosed by AI. The 30-day mortality rates were 74.5% (38/51) in type A and 25.0% (7/28) in type B. The age-adjusted incidence and mortality of AAD per 100 000 inhabitants were 17.6 (type A: 11.3, type B: 6.2) and 9.9 (type A: 8.4, type B: 1.5), respectively. The population-based survey of AAD showed that the age-adjusted incidence of AAD was two-fold higher than in previous reports, and the actual mortality rates were markedly higher due to the high incidence of dead-on-arrival.

Fatal Case of Dysphagia Induced by Neurofibromatosis Type II

A 17-year-old boy was found in an unconscious state. He had had neurofibromatosis type 2 (NF2) since 9 years old. His history included multiple operations for NF2 at the left jugular vein foramen, thoracic spinal cord, and cervical spinal cord at 9 and 10 years old. However, remaining bilateral acoustic tumors resulted in the onset of aspirated pneumonia within a few months. He was found unconscious in his dormitory. When emergency medical technicians checked him, he was in cardiac arrest. Saburra was identified in his mouth. He did not achieve return of spontaneous circulation with advanced cardiac life support. Autopsy imaging revealed residual bilateral acoustic tumors compressing the brain stem. NF2 is a serious disease that can rapidly become life-threatening. We hope new treatments that will improve the survival of NF2 patients will be developed soon.

Radiologically Undetectable Latent Pulmonary Lymphocyte Infiltration Suggesting the Need for Early Glucocorticoid Therapy: An Autopsy Case of Rapidly Progressive COVID-19 Pneumonia with Lymphopenia.

We herein report a fatal case of coronavirus disease 2019 (COVID-19) pneumonia with rapid progression of respiratory failure and lymphopenia. Excessive recruitment and sequestration of lymphocytes in the lung were suggested as the pathophysiology underlying COVID-19-associated lymphopenia. Interestingly, the autopsy in this case revealed lymphocytic infiltration in the lungs even at sites that appeared normal on autopsy imaging. These findings suggest that in COVID-19 cases with risk factors of severe exacerbation, early glucocorticoid administration should be considered, especially if lymphopenia is present, even if the imaging findings show only mild abnormalities.

Postmortem Computed Tomography Scan as Autopsy Imaging at a Single Tertiary Center：1-Year Review

Postmortem Computed Tomography Scan as Autopsy Imaging at a Single Tertiary Center：1-Year Review

Autopsy imaging（Ai）により詳細診断をした人魚体の一例

Autopsy imaging（Ai）により詳細診断をした人魚体の一例

Autopsy imaging for aspergillosis in King Penguin, an economically valuable zoo animal.

Autopsy imaging (Ai) was performed for a King Penguin. Ai-computed tomography (CT) revealed air sac membrane thickening, multiple nodules in the cranial air sac, suspected abscess, lung infiltration, and air sac contraction. Based on these findings, respiratory disorder was concerned. Aspergillosis, which is the highly observed in penguins, was considered as the primary differential diagnosis. The cultured sample showed characteristic conidial head of Aspergillus spp., the DNA of which was 100% identical to that of A. fumigatus. The cause of death was determined to respiratory failure due to aspergillosis. Ai-CT findings facilitated the dissection workflow and alerted the pathologist to potential hazards during the autopsy. Ai is useful to determine the cause of death and for readiness and safe pathological dissection.

Non-contrast computed tomography of type A acute aortic dissection in patients with out-of-hospital cardiopulmonary arrest: a case series

BackgroundThe prognosis of patients admitted for acute aortic dissection (AAD) has remarkably improved. However, we must also consider out-of-hospital cardiopulmonary arrest (OHCPA) patients while assessing the prognosis. In recent years, autopsy imaging has become more common as an alternative to conventional autopsy. Therefore, we reviewed our OHCPA patients with type A AAD using acute phase non-contrast computed tomography (CT).Case summaryHere, we report a case series of three patients who developed OHCPA and were diagnosed with type A AAD using non-contrast CT. Although the direct causes of death varied in each case, we could easily determine the direct causes of death from clinical course of the condition and from non-contrast CT.DiscussionAlthough non-contrast CT does not completely replace autopsy, if its convenience and non-invasiveness make it possible for more patients to undergo the procedure, the real prognosis (including morbidity and mortality) may be better understood. Therefore, we considered it significant to use non-contrast CT for investigating the cause of sudden death.

Autopsy Imaging and Medical Accident Investigation System

Autopsy Imaging and Medical Accident Investigation System

Workshop on Autopsy Imaging in Gunma Prefecture

Workshop on Autopsy Imaging in Gunma Prefecture

The significance of the autopsy imaging (Ai) in Japanese corpse examination

The significance of the autopsy imaging (Ai) in Japanese corpse examination

Usefulness of autopsy imaging for cases of death after out-of-hospital cardiopulmonary arrest

Usefulness of autopsy imaging for cases of death after out-of-hospital cardiopulmonary arrest

Postmortem interval estimation using the animal model of postmortem gas volume changes

It is important to estimate the postmortem interval in forensic autopsy. Many methods to estimate the postmortem interval have been reported, and are typically associated with internal examination. However, there are issues such as rejection of autopsy by the family and a lack of forensic doctor in internal examination. Therefore, it is necessary to develop new methods, such as autopsy imaging, that can substitute for internal examination. Here, we first evaluated whether gas volume in the body increased with postmortem interval. Time-dependent X-ray CT imaging of euthanized Crl:CD (SD) rats (n = 3) was performed immediately after euthanasia and at seven subsequent time points up to 168 h (7 days) at 24-hour intervals. The data revealed that gas volume in the body increased in a time-dependent manner. Next, we reconstructed 3D images of isolated gas and calculated the gas volume using Amira software. In all cases, the volume of both portal venous gas and intestinal gas increased in a time-dependent manner. The volume of portal venous gas increased exponentially, while the volume of intestinal gas increased in a linearly with time. These data might be suggested that the postmortem gas volume change is one of indicators for estimating the postmortem interval. In addition, it would be possible to estimate more accurate postmortem interval by combining not only gas volume changes at the above two sites but also gas volume changes of the other sites such as heart cavities, kidney parenchyma, or abdominal aorta.