Verification of the effect of data-driven brain motion correction on PET imaging.

INTRODUCTION: Brain positron emission tomography/computed tomography (PET/CT) scans are useful for identifying the cause of dementia by evaluating glucose metabolism in the brain with F-18-fluorodeoxyglucose or Aß deposition with F-18-florbetaben. However, since imaging time ranges from 10 to 30 minutes, movements during the examination might result in image artifacts, which interfere with diagnosis. To solve this problem, data-driven brain motion correction (DDBMC) techniques are capable of performing motion corrected reconstruction using highly accurate motion estimates with high temporal resolution. In this study, we investigated the effectiveness of DDBMC techniques on PET/CT images using a Hoffman phantom, involving continuous rotational and tilting motion, each expanded up to approximately 20 degrees. MATERIALS AND METHODS: Listmode imaging was performed using a Hoffman phantom that reproduced rotational and tilting motions of the head. Brain motion correction processing was performed on the obtained data. Reconstructed images with and without brain motion correction processing were compared. Visual evaluations by a nuclear medicine specialist and quantitative parameters of images with correction and reference still images were compared. RESULTS: Normalized Mean Squared Error (NMSE) results demonstrated the effectiveness of DDBMC in compensating for rotational and tilting motions during PET imaging. In Cases 1 and 2 involving rotational motion, NMSE decreased from 0.15-0.2 to approximately 0.01 with DDBMC, indicating a substantial reduction in differences from the reference image across various brain regions. In the Structural Similarity Index (SSIM), DDBMC improved it to above 0.96 Contrast assessment revealed notable improvements with DDBMC. In continuous rotational motion, % contrast increased from 42.4% to 73.5%, In tilting motion, % contrast increased from 52.3% to 64.5%, eliminating significant differences from the static reference image. These findings underscore the efficacy of DDBMC in enhancing image contrast and minimizing motion induced variations across different motion scenarios. CONCLUSIONS: DDBMC processing can effectively compensate for continuous rotational and tilting motion of the head during PET, with motion angles of approximately 20 degrees. However, a significant limitation of this study is the exclusive validation of the proposed method using a Hoffman phantom; its applicability to the human brain has not been investigated. Further research involving human subjects is necessary to assess the generalizability and reliability of the presented motion correction technique in real clinical scenarios.

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.

A case report of forensic personal identification by transposed teeth and dental treatment marks using multidetector row computed tomography.

Using postmortem computed tomography (CT) images, we achieved personal identification of a body using transposed teeth and dental treatment marks. Transposition of teeth is a rare malpositioning anomaly. CT images can clarify the malpositioning of the teeth's roots, which is difficult to discern from gross observation of the dentition. Because dental hygiene has reduced the incidence of caries in recent years, it might be difficult to use treatment marks for personal identification in the future. Transposed teeth, although rare, provide an important clue to personal identification.

Deep Learning-Based Diagnosis of Fatal Hypothermia Using Post-Mortem Computed Tomography.

In forensic medicine, fatal hypothermia diagnosis is not always easy because findings are not specific, especially if traumatized. Post-mortem computed tomography (PMCT) is a useful adjunct to the cause-of-death diagnosis and some qualitative image character analysis, such as diffuse hyperaeration with decreased vascularity or pulmonary emphysema, have also been utilized for fatal hypothermia. However, it is challenging for inexperienced forensic pathologists to recognize the subtle differences of fatal hypothermia in PMCT images. In this study, we developed a deep learning-based diagnosis system for fatal hypothermia and explored the possibility of being an alternative diagnostic for forensic pathologists. An in-house dataset of forensic autopsy proven samples was used for the development and performance evaluation of the deep learning system. We used the area under the receiver operating characteristic curve (AUC) of the system for evaluation, and a human-expert comparable AUC value of 0.905, sensitivity of 0.948, and specificity of 0.741 were achieved. The experimental results clearly demonstrated the usefulness and feasibility of the deep learning system for fatal hypothermia diagnosis.

Improvement of Imaging Conditions to Improve the Detection Rate of Head and Neck Cancer by Positron Emission Tomography/Computed Tomography Examination.

Positron emission tomography (PET)/computed tomography (CT) has improved sensitivity and resolution using silicon photomultiplier as a photosensor. Previously, only a fixed setting was available for the shooting time of 1 bed, but now, the shooting time can be changed for each bed. Time can be shortened or extended depending on the target area. A few studies reported on image reconstruction conditions for head and neck cancer in whole-body PET/CT examinations. Thus, this study aimed to optimize the imaging conditions of the head and neck region during whole-body imaging. A cylindrical acrylic container with a 200 mm diameter was used to simulate the head and neck area using a PET/CT system equipped with a semiconductor detector. Spheres of 6-30 mm in diameter were enclosed in the 200 mm diameter cylindrical acrylic vessel. Radioactivity in 18F solution (Hot:BG ratio 4:1) was enclosed in a phantom following the Japanese Society of Nuclear Medicine (JSNM) guidelines. Background radioactivity concentration was 2.53 kBq/mL. List mode acquisition of 1,800 s was collected at 60-1,800 s with the field of view of 700 mm and 350 mm. The image was reconstructed by resizing the matrix to 128 × 128, 192 × 192, 256 × 256, and 384 × 384, respectively. The imaging time per bed in the head and neck should be at least 180 s, and the reconstruction conditions should be a field of view (FOV) of 350 mm, matrix sizes of ≥ 192, and a Bayesian penalized likelihood (BPL) reconstruction with a ß-value of 200. This allows detection of > 70% of the 8-mm spheres in the images.

A 2.5D Deep Learning-Based Method for Drowning Diagnosis Using Post-Mortem Computed Tomography.

It is challenging to diagnose drowning in autopsy even with the help of post-mortem multi-slice computed tomography (MSCT) due to the complex pathophysiology and the shortage of forensic specialists equipped with radiology knowledge. Therefore, a computer-aided diagnosis (CAD) system was developed to help with diagnosis. Most deep learning-based CAD systems only utilize 2D information, which is proper for 2D data such as chest X-ray images. However, 3D information should also be considered for 3D data like CT. Conventional 3D methods require a huge amount of data and computational cost when using 3D methods. In this article, we proposed a 2.5D method that converts 3D data into 2D images to train 2D deep learning models for drowning diagnosis. The key point of this 2.5D method is that it uses a subset to represent the whole case, covering this case as much as possible while avoiding other repetitive information. To evaluate the effectiveness of the proposed method, conventional 2D, previous 2.5D, and 3D deep learning-based methods were tested using an MSCT dataset obtained from Tohoku university. Then, to provide explainable diagnosis results, a visualization method called Gradient-weighted Class Activation Mapping was employed to visualize features relevant to drowning in CT images. Results on drowning diagnosis showed that our proposed method achieved the best performance compared to other 2D, 2.5D, and 3D methods. The visual assessment also demonstrated that our method could find the saliency regions corresponding to drowning.

Diagnosing Drowning in Postmortem CT Images Using Artificial Intelligence.

Imaging features of the lung in postmortem computed tomography (CT) scans have been reported in drowning cases. However, it is difficult for forensic pathologists with limited experience to distinguish subtle differences in CT images. In this study, artificial intelligence (AI) with deep learning capability was used to diagnose drowning in postmortem CT images, and its performance was evaluated. The samples consisted of high-resolution CT images of the chest of 153 drowned and 160 non-drowned bodies captured by an 8- or 64-row multislice CT system. The images were captured with an image slice thickness of 1.0 mm and spacing of 30 mm, and 28 images were typically captured. A modified AlexNet was used as the AI architecture. The output result was the drowning probability for each component image. To evaluate the performance of the proposed model, the area under the receiver operating characteristic curve (AUC) was analyzed, and the AUC value of 0.95 was obtained. This indicates that the proposed AI architecture is a useful and powerful complementary testing approach for diagnosing drowning in postmortem CT images. Notably, the accuracy was 81% (62/77) for cases in which resuscitation was performed, and 92% (216/236) for cases in which resuscitation was not attempted. Therefore, the proposed AI method should not be used to diagnose the cause of death when aggressive cardiopulmonary resuscitation was performed. Additionally, because honeycomb lungs are likely to exhibit different morphologies, emphysema cases should also be treated with caution when the proposed AI method is used to diagnose drowning.

Diagnostic Accuracy of Liver Damage Based on Postmortem Computed Tomography Findings in High-Energy Trauma.

The liver is an organ that is frequently injured by blunt trauma. In clinical medicine, contrast-enhanced computed tomography (CT) is useful for diagnosing liver damage. However, detection of liver injury is difficult with postmortem CT (PMCT) that does not use contrast media. This study aims to identify findings that are useful for diagnosing liver injury with PMCT. This study included 97 high-energy trauma cases that underwent both PMCT without contrast and forensic autopsy between July 2009 and November 2020. PMCT findings in and around the liver in cases of liver injury were collected. The diagnostic accuracy of each finding was calculated. Of 97 cases, 62 had liver injury on autopsy. PMCT detected hepatic surface gas in 31 cases, intrahepatic focal gas in 19 cases, high-density fluid accumulation around the liver in 12 cases, and fracture of a right lower (9th-12th) rib in 48 cases. Abnormal PMCT findings in and around the liver had insufficient diagnostic sensitivity, ranging from 19.4% to 38.7%. By contrast, the finding of a right lower rib fracture was more sensitive for suspected liver injury. Our results indicate that abdominal liver findings (hepatic surface gas, intrahepatic focal gas, and high-density fluid accumulation around the liver) are of limited utility for the diagnosis of liver injury, whereas a right lower rib fracture is a useful indirect finding.

Cervical intervertebral separation caused by trauma on post-mortem computed tomography: Possibility of a diagnosis based on intervertebral gas.

OBJECTIVES: Gas is a common finding in cervical intervertebral separation. However, intervertebral gas is also found in many decedents without intervertebral separation. Here, we quantified intervertebral gas and examined its value in the diagnosis of cervical intervertebral separation. METHODS: We retrospectively reviewed data from 1118 decedents who underwent post-mortem computed tomography (CT) and autopsy from May 2011 to July 2016 and selected those with cervical intervertebral gas with or without intervertebral separation. These data comprised 56 cervical intervertebral spaces with gas [intervertebral separation in 19 (33.9%)] in 43 subjects [intervertebral separation in 17 (39.5%)]. We categorised the decedents according to gas volume, position, and shape and determined the significance of the differences between the decedents with and without separation. RESULTS: The gas volume did not differ significantly between decedents with and without separation (p = 0.063). However, there were significant differences in the gas position between decedents with and without separation. In the sagittal plane: gas was seen in the "centred" position in the ventral-to-dorsal direction in more decedents without separation than in those with separation (p = 0.018). Gas was seen in the ventral-to-dorsal positions in more decedents with separation than in those without separation (p = 0.049). In the cranio-caudal direction, gas in the upper position was more common in decedents with separation than in those without separation in the sagittal plane (p = 0.03). In the coronal plane: gas was seen in the upper position more frequently in decedents with separation in the cranio-caudal direction than in those without separation (p = 0.001). A significant difference in gas shape was observed only in the coronal plane (p = 0.024); irregular gas was associated with decedents without separation. CONCLUSION: Gas in the ventral-to-dorsal and upper positions in the sagittal plane and in the upper position in the coronal plane was rather indicative of cervical intervertebral separation. An irregular gas shape in the coronal plane was indicative of degenerative changes.

Block-like and cast-like hyperdense areas in the right heart cavities on post-mortem CT strongly suggest the presence of intracardiac blood clots at autopsy.

OBJECTIVE: To classify the types of hyperdense areas in the heart cavities on post-mortem CT (PMCT) and compare them according to the presence of blood clots in the heart cavities at forensic autopsy. METHODS: One hundred and twelve cases with CT images taken before forensic autopsy were evaluated. The presence and shape of hyperdense areas in the right or left heart cavities were retrospectively evaluated on PMCT images and were classified into four types (block-like, cast-like, fluid level-like, and unclear). The presence of blood clots was confirmed when there were clots in the heart cavities at forensic autopsy. RESULTS: Of the 112 cases, 57 exhibited blood clots in the heart cavities at forensic autopsy. The hyperdense areas in the right heart cavities on PMCT in 57 cases exhibiting blood clots at forensic autopsy were classified as follows: block-like, 32; fluid level-like, 4; cast-like, 17; and unclear, 4. The sensitivity of block-like and cast-like hyperdense areas in the right heart cavities on PMCT for the presence of clots in the heart cavities at forensic autopsy was 86% (95% confidence interval [CI]: 74-94%); the corresponding specificity, PPV, and NPV were 95% (95% CI: 85-99%), 94% (95% CI: 84-99%), and 87% (95% CI: 75-94%), respectively. CONCLUSION: Block-like and cast-like hyperdense areas in the right heart cavities on PMCT predicted the presence of intracardiac blood clots at forensic autopsy. KEY POINTS: • Clinical radiologists likely have no experience of interpreting findings of blood clots on post-mortem CT (PMCT). • The appearance of blood clots on PMCT provides important clues for diagnosing the cause and process of death. • The shapes of the hyperdense areas in the heart cavities were classified into four types, and two of these types could be used to predict the presence of blood clots in the heart cavities at forensic autopsy.

A Deep Learning Aided Drowning Diagnosis for Forensic Investigations using Post-Mortem Lung CT Images.

Feasibility of computer-aided diagnosis (CAD) systems has been demonstrated in the field of medical image diagnosis. Especially, deep learning based CAD systems showed high performance thanks to its capability of image recognition. However, there is no CAD system developed for post-mortem imaging diagnosis and thus it is still unclear if the CAD system is effective for this purpose. Particulally, the drowning diagnosis is one of the most difficult tasks in the field of forensic medicine because findings of the post-mortem image diagnosis are not specific. To address this issue, we develop a CAD system consisting of a deep convolution neural network (DCNN) to classify post-mortem lung computed tomography (CT) images into two categories of drowning and non-drowning cases. The DCNN was trained by means of transfer learning and performance evaluation was conducted by 10-fold cross validation using 140 drowning cases and 140 non-drowning cases of the CT images. The area under the receiver operating characteristic curve (AUC-ROC) for the DCNN was achieved 0.88 in average. This high performance clearly demonstrated that the proposed DCNN based CAD system has a potential for post-mortem image diagnosis of drowning.

[Examination of Optimal Window Size and Acquisition Time of Respiratory-gated PET Image: Phantom Study with a SiPM-based PET/CT Scanner].

PURPOSE: This phantom study aimed to determine the optimal acquisition window size for phase-based respiratory gating in silicon photomultiplier (SiPM)-based fluorodeoxyglucose positron emission tomography/computed tomography (FDG-PET/CT) and its acquisition time in respiratory-gated imaging with the optimal window size. METHODS: Images of a moving NEMA IEC Body Phantom SetTM with hot spheres were acquired. First, the tumor volume and the maximum standardized uptake value (SUVmax) of images reconstructed using a different window size were evaluated to define the optimal window size. Second, the quality of the images reconstructed using the optimal window size and different acquisition times was evaluated using the detectability score of the 10-mm hot sphere and physical indices. RESULTS: The volume and the SUVmax of the 10-mm hot sphere were improved when the window size was narrow, and there were no significant differences among images reconstructed using a window size narrower than 20%. To reconstruct an image using the 20% window size, an acquisition time of 5 min was required to visualize the 10-mm hot sphere. CONCLUSIONS: The optimal window size for phase-based respiratory gating is 20%. Further, an acquisition time of 5 min should be taken for respiratory-gated imaging with the 20% window size on SiPM-based FDG-PET/CT.

A quantitative morphological analysis of three-dimensional CT coxal bone images of contemporary Japanese using homologous models for sex and age estimation.

Sexual dimorphisms and age-dependent morphological features of the human coxal bone were quantitatively analyzed using homologous models created from three-dimensional (3D) computed tomography images of the pelvis (male: 514 samples, female: 388 samples, age 16-100). Bilateral average coxal images of each sex and age decade were generated separately through principle component analyses (PCA). By measuring average point-to-point distances of 8472 corresponding points (average corresponding point differences [ACPDs]) between each homologous coxal image and the average images, the sex of more than 93% of the samples was correctly assigned. Some principal components (PCs) detected in PCA of the homologous models of the samples correlated fairly well with age and are affected by features of the curvature of the iliac crest, the arcuate line and the greater sciatic notch. Moreover, separate PCA using the average images of each age decade successfully detected the first PCs, which were strongly correlated with age. However, neither multiple regression analysis using PCs related to age nor comparison of ACPDs with the average images of each age decade could produce accurate results for age decade assignment of unknown (blind) samples. Therefore, more detailed analysis of age-dependent morphological features would be necessary for actual age estimation. In addition, some laterality or left and right shape difference of the coxal bone images was also elucidated, and was more significant in females. Analysis of 3D structures using homologous models and PCA appears to be a potential technique to detect subsistent morphological changes of bones.

Sudden death in a child caused by a giant cavernous hemangioma of the anterior mediastinum.

A 4-year-old girl who had been treated for asthma since the age of 2 years had a severe coughing fit and died suddenly. The patient had a history of occasional severe coughing fits, and these fits had been worsening in severity during the week before her death. Prior to death, she was taken to a clinician, and thymic hypertrophy was suspected based on chest X-ray findings. The clinician recommended that she visit a general hospital at a later date; however, she died that night. Postmortem radiology and autopsy revealed a large mass in the anterior mediastinum compressing the heart and airway, and no other findings attributable to sudden death were observed. Therefore, we concluded that the patient's death was attributable to acute respiratory and cardiac circulatory failure secondary to the pressure induced by the mass. Microscopically, the mass showed a cavernous structure composed of cystically dilated, thin-walled large vessels filled with blood. The final diagnosis was a cavernous hemangioma. Hemangiomas are the most common benign vascular anomalies seen in young children; however, mediastinal hemangiomas are rare and can cause life-threatening complications because of their size and location. Therefore, forensic pathologists should include hemangioma as a differential diagnosis in children with anterior mediastinal masses.

Postmortem Computed Tomographic Analysis of Death Caused by Oral Drug Intoxication.

Traditional autopsy has changed little in the past century. In Japan, the rate of forensic autopsy in cases of unusual death is very low. Therefore, multi-slice computed tomography (CT) has been used to obtain imaging data instead of or in addition to autopsy in suspicious forensic cases. In our institute, postmortem multi-slice CT has been performed since 2009, and by 2014 there were over 1,000 cases. Our extensive experience with postmortem CT shows that in many cases of death by drug overdose, stomach contents exhibit high X-ray absorption. This article reviews the relationship between CT findings of stomach contents and toxicological analysis results in 23 cases of death by drug overdose. All cases (12 females and 11 males, aged 44 ± 11 years) known to have orally ingested drugs were included in this study. We assessed the slices of all stomach areas on consecutive axial CT images. Twenty cases (87%) showed high X-ray absorption in the stomach, while the other three did not demonstrate radio-dense stomach contents even though drug analysis detected lethal concentrations of drugs in the blood. In conclusion, drugs were frequently, but not always, visualized as contents with high X-ray absorption in the stomach. Postmortem gastric CT images can provide useful information in cases of oral drug intoxication if there are empty drug packages or a suicide note at the death scene. However, precise determination of the cause of death requires full autopsy in cases where there is no indication of suicide at the death scene.

Age estimation by ossification of thyroid cartilage of Japanese males using Bayesian analysis of postmortem CT images.

Many studies have reported that the degree of thyroid cartilage ossification correlates with age. However, its use in forensics has been limited by individual variation, tissue fragility, and observer subjectivity. We examined a new method for age estimation from the degree of ossification of thyroid cartilage using postmortem computed tomography (CT) and Bayesian analysis. The scoring of CT density values, ossification rate, and morphological changes was performed in 131 Japanese males (20-80years old), and an age estimation table was created and verified in 39 other Japanese males. We tried two types of prior probability; one was males undergoing postmortem CT imaging at our institute, and the other was the age composition of the male population in Japan. The differences in percentage in each age group were observed, especially in the ninth decade, but they were not large enough to change the result. Two decedents in their 30s and 80s could not be assigned an age group. Likelihood might have had more significant influence than prior probability. Bayesian approach and new scoring system of thyroid cartilage ossification might be useful as an age estimation method. Although the low posterior probability in decedents the age ≧70 remains a problem, the advantage of Bayesian estimation that it can sequentially update the age estimation table by factoring in exceptional data can solve this problem. In this study, accuracy of age estimation did not dramatically increase, however, Bayesian statistics might become a useful technique for examining any tissue samples subject to age-related calcification.

A Case Report of Postmortem Radiography of Acute, Fatal Abdominal Distension After Binge Eating.

This case report describes a woman who developed fatal gastric dilatation after binge eating. She called an ambulance because of stomach pain. When she arrived at the hospital, she did not look seriously ill. However, she rapidly became unconscious and collapsed immediately after she was laid on the examination table in a supine position. Postmortem chest x-ray and computed tomography showed right shift of the mediastinum and raised left diaphragm caused by massive gastric distension. Computed tomography showed no visible inferior vena cava. We think that her sudden deterioration was caused by movement of her stomach contents. Radiographic images provided some clues to the cause of her rapid collapse and death.

A case of fatal sigmoid volvulus visualized on postmortem radiography: The importance of image optimization with multidetector computed tomography.

This report describes the case of a man who developed fatal sigmoid volvulus that was identified on postmortem radiography before forensic autopsy. Postmortem radiography is useful for visualizing the body prior to autopsy. We discuss postmortem multidetector computed tomography that was tailored for optimum image quality to allow reconstruction of the fatal findings in multiple axes and in three dimensions, helping to pinpoint the anatomical sites of interest. This involves techniques such as manipulation of the scanning beam pitch and overlapping CT section acquisition. These techniques are best performed by personnel with CT technology training.

Usefulness and limitations of postmortem computed tomography in forensic analysis of gunshot injuries: Three case reports.

Gunshot injury has always been an important field of investigation in postmortem forensic radiology. The localization and retrieval of the bullet and of potentially important fragments are vital to these cases. Using postmortem multidetector-row computed tomography (MDCT) prior to forensic autopsy, we sought to illustrate the importance of this modality in the noninvasive characterization of gunshot wounds. We obtained and analyzed MDCT images in three cases of gunshot wounds (accidental close-range shotgun shooting, suicidal contact gunshot to the head and accidental long-range buckshot shooting). We discuss the value of postmortem MDCT findings in gunshot wound cases by comparing with forensic autopsy findings in Japan, a developing country with miserably low autopsy rate.