Journal of Forensic Radiology and Imaging最新文献

Objective: Fields such as biological anthropology, forensic anthropology, archaeology, and mummy studies have increased the use medical imaging, such as computed tomography (CT), to analyze human remains through virtual examination. When predicting chronological age from virtual anatomy, methods developed on skeletal collections do not always produce accurate age predictions when applied to virtual anatomy. Age estimation methods developed specifically for medical imaging can improve accuracy of chronological age prediction when examining virtual anatomy. The present study examines the relationship between attenuation of trabecular bone from the os coxae and chronological age using CT scanning data.

Materials and Methods: A sample of 324 CT scans of living and cadaveric individuals were randomly selected from a CT scan database and used to identify a correlation between trabecular bone CT attenuation in four regions of the os coxae and chronological age. The four regions included trabecular bone deep to the auricular surface, pubic symphysis, posterior superior iliac spine, and the ischial tuberosity.

Results: Through cross validation, the trabecular bone deep to the pubic symphysis, posterior superior iliac spine, and ischial tuberosity resulted in the best prediction models based on model testing. Model testing identified models with adjusted R2 values of 0.83 and 0.86 for combine male/female bone attenuation and only female bone density, respectively.

Discussion: The resulting models, when used in relation with CT scanning data, provide a quantitative method that predicts chronological age and can be used in situations when remains are recently deceased fleshed individuals.

This work proposes a novel approach to estimate the sensitivity of post-mortem computed tomography (PMCT) with respect to traditional necropsy (gold standard). Using concepts of Bayes theorem and the opinion of 57 veterinarians of how a relevant finding (RF) on PMCT or at necropsy can be a determinant in the animal's death, the general sensitivity was estimated. Such sensitivity can be interpreted as the probability to identify a specific RF in a necropsied animal. In the study, we included 18 animals that were subjected to traditional necropsy and underwent PMCT, which provided 48 different findings; all were assumed as a potential cause of death. This study aimed to understand the probability of death, given that a necropsied animal presented a specific RF. To estimate all contributions of the 48 RF, we designed a simple survey based on the Delphi method. In this survey, we invited veterinarians to evaluate the 48 NF and to assign a grade in a Likert scale from 0 to 10 for each of the RFs. In this scale, 0 (zero) indicates that the NF does not contribute to the animal's death, and 10 (ten) indicates that this NFdoes totally contributes to the animal's death. Therefore, each veterinarian, according to their own professional experience, assigns a grade describing how important each necroscopic finding is to the animal's death. Our results demonstrated that the problem of sensitivity analysis in the absence of a gold standard could be circumvented by the method proposed here.

Objective

When present, the most commonly used element to estimate adult skeletal age at death is the pubic symphysis. In the second manuscript of this series, the Suchey–Brooks and Hartnett pubic symphysis methods are assessed to for their applicability to volume-rendered images (VRIs) generated from CT scans of cadavers to assess their applicability to VRIs.

Materials and Methods

CT scans of 420 cadavers from the Victorian Institute of Forensic Medicine were selected (age range 20 to 79 years). Siemens syngo.via software was used to view the DICOM images and create the VRIs. The Suchey–Brooks and Hartnett pubic symphysis methods were used to create a revised method to estimate age for VRIs.

Results and Conclusion

Several features described by Suchey–Brooks and Hartnett could be observed on the VRIs, including billows on the symphysis face in younger individuals, the ventral rampart, and the dorsal margin. However, distinguishing features such porosity and bone weight for the older phases could not always be assessed. A revised pubic symphysis method for use on VRIs was created using both morphological features and bone quality to assess age. The revised method is strongly correlated with age, and over 70% of the sample was placed in the correct age phase. The Suchey–Brooks method was more accurate for individuals under 40 years of age while the revised method was more accurate for individuals over 40 years of age, similar to the findings from other studies. This revised method should be tested on other populations using different CT settings and software.

Objectives

Postmortem computed tomography (PMCT) usually includes the generation of great amounts of imaging data, and is often reviewed by forensic pathologists. To allow a more resource-efficient diagnosis, deep neural networks may act as a pre-scanning tool in postmortem radiology. In this study, a deep neural network to classify cases depending on the presence skull fractures on curved maximum intensity projections (CMIP).

Methods

Calvarial CMIPs of each 75 cases with and without documented skull fractures were retrospectively generated from our database. Then, half of the data were randomly assigned to either training or validation. In supervised training, fractures were manually marked. During validation, each image received a gradual score between 0 and 1 predicting the likelihood of showing one or more fractures.

Results

With a total number of 100 networks trained, the average area under the Receiver Operating Characteristic curve (AUC) was 0.895. The best performing network had an AUC of 0.965. At a classification threshold of 0.79, the network classified fracture cases correctly with a sensitivity of 91.4% and a specificity of 87.5%.

Conclusion

Classification based on the existence of skull fractures on CMIPs with deep learning is feasible. For the purpose of pre-scanning PMCT data, a classification threshold of 0.75 with a sensitivity of 100% can be applied. A higher number of images of validated skull fractures available will increase the performance of the network. In the future, Deep learning might enable a more resource-efficient assessment in postmortem radiology.

The present study was designed to measure and compare various measurements of orbital apertures between male and female subjects using digital computed tomography. Also to assess the usefulness of orbital aperture measures as an aid in sex determination in sample of Egyptian population. The study included 92 subjects (44 males and 48 females) with age ranged from 18 to 65 years. The results showed that males exhibited greater mean values for all the measurements except for left orbital width that was slightly increased in females. There were significant differences (p < 0.05) between males and females regarding right orbital height, right and left area and inter zygomatic distance. The accuracy of correct sex classification was achieved up to 74.7%. Discriminant analysis revealed that inter zygomatic distances showed significant discrimination (p = 0.028) between both sexes. Receiver operating characteristic (ROC) curve analysis for sex discrimination demonstrated that inter zygomatic distance at a cutoff of 96.7, had sensitivity of 76.7%, specificity of 65% and accuracy 71% (p < 0.001). In conclusion, inter zygomatic distance was found to be the best reliable parameter and achieved the highest accuracy 71%. So, it was seen that inter zygomatic distance could be used for the sex determination from computed tomography.

The present study aimed to analyze morphometric variables of the mastoid process using X-ray microtomography images for sex estimation. The sample was composed of 60 microtomography images from dried skulls. Three measurements were applied: asterion-porion, mastoideale-porion and asterion-mastoideale. The area of the triangle was calculated by two methodologies: software myVGL2.2 + Heron's formula (M1) and software myVGL2.2 + Adobe Photoshop CS6 (M2). The right area (RA) and left area (LA) were calculated, and the sum of them was considered the total area (TA). Descriptive analysis of M1 and M2 showed statistically significant differences between sex (p < 0.05). The univariate discriminant analysis showed an overall percentage of correct classification of sex varying between 75–80% for M1 and 71.7–73.3% for M2. The multivariate discriminant analysis demonstrated reduced classification percentages. The ROC curve analysis showed better results for M1 compared to M2. The best variable using M1 was the TA, with an 81.45% accurate sex estimation. Based on the ROC curve analysis results, a reference table was developed to be applied for the sex estimation of Brazilian individuals. In conclusion, the mastoid process has good sexual dimorphism in Brazilian skulls and can be considered a tool for predicting sex in cases of fragmented craniums.

Objective

CT scans have become a standard part of autopsy procedures in large medical-legal facilities, and forensic anthropologists are increasingly asked to contribute their knowledge of skeletal anatomy to volume-rendered images (VRIs) generated from CT scans. However, it is not clear if the age estimation methods created using skeletal remains can be reliably applied to VRIs. This study, along with the other two in this series, examines the applicability of three commonly used sites for adult skeletal age estimation on VRIs generated from CT scans of cadavers. Part I focuses on the fourth rib.

Materials and methods

CT scans of 420 cadavers from the Victorian Institute of Forensic Medicine were selected (age range 20–79 years). Siemens syngo.via software was used to view the DICOM images and create the VRIs. The İşcan et al. and Hartnett fourth rib methods of age estimation were used to create a revised method for VRIs.

Results and conclusion

Several features described by both İşcan et al. and Hartnett could be observed on the VRIs; however, features such as porosity inside the sternal rib and bone weight could not be assessed. A revised fourth rib method for use on VRIs was created. Significantly more individuals were placed in the correct age phase using the revised method compared to the İşcan et al. and Hartnett methods, and overall the revised method had significantly improved inaccuracy and bias scores. The revised method is reliable and should be tested on other populations using different software.

Objective

The auricular surface of the ilium is often found preserved in both archaeological and forensic contexts. In the final manuscript of this three-part series, the features used by biological and forensic anthropologists to estimate adult skeletal age from the auricular surface are tested on volume-rendered images (VRIs) generated from CT scans of cadavers.

Materials and methods

CT scans of 420 cadavers from the Victorian Institute of Forensic Medicine were selected (age range 20–79 years). Siemens syngo.via software was used to view the DICOM images and create the VRIs. The Lovejoy et al. auricular surface method of age estimation was assessed for use on VRIs and a new method for VRIs was created.

Results and conclusion

Most of the features described by Lovejoy et al. could not be observed on the VRIs, and as previously established by Villa et al. (2013) [13], the Buckberry and Chamberlain method cannot be used on VRIs. Features such as transverse organization, surface texture, and microporosity are not visible on VRIs; however, features of the auricular surface and retroauricular area were seen to progress in a manner similar to that described by Lovejoy and colleagues, and a revised auricular surface method for use on VRIs was developed. The revised method is strongly correlated with age, and over 70% of the sample was placed within one standard deviation of the mean of the correct phase. This revised method should be tested on more populations using different CT settings and software.