Image segmentation of post-mortem computed tomography data in forensic imaging: Methods and applications

[Features of detection and interpretation of intravital and postmortem changes according to the results of traditional X-ray and X-ray computed tomography of objects from historical graves and artefacts]

OBJECTIVE

To study emergence mechanism, physical nature, pattern of intravital and postmortem changes of biological and non-biological objects originated in the period from 1550 to 1918 yr. using traditional X-ray and X-ray computed tomography.

MATERIAL AND METHODS

The relics of Saint Macarius the Roman of Novgorod, the remains of the First Reverend of the Resurrection Novodevichy Convent in Saint Petersburg Mother Superior Theophania, damages on the chair leg on which Tsesarevich Alexey sat during the shooting of Russian Emperor Nicholas II, his family and entourage in 1918 in Yekaterinburg were stidued.

RESULTS AND CONCLUSION

The application of highly informative methods of traditional X-ray and X-ray computed tomography of biological and non-biological objects showed their high informativity and allowed to correctly interpret the emergence mechanism, physical nature, pattern of intravital and postmortem changes of skeleton bones and historical artefact (chair legs) originated long ago. The necessity of special professional training and advanced training of experts in forensic radiology to prevent possible diagnostic and expert errors has been substantiated.

A Virtual, 3D Multimodal Approach to Victim and Crime Scene Reconstruction

In the last two decades, forensic pathology and crime scene investigations have seen a rapid increase in examination tools due to the implementation of several imaging techniques, e.g., CT and MR scanning, surface scanning and photogrammetry. These tools encompass relatively simple visualization tools to powerful instruments for performing virtual 3D crime scene reconstructions. A multi-modality and multiscale approach to a crime scene, where 3D models of victims and the crime scene are combined, offers several advantages. A permanent documentation of all evidence in a single 3D environment can be used during the investigation phases (e.g., for testing hypotheses) or during the court procedures (e.g., to visualize the scene and the victim in a more intuitive manner). Advanced computational approaches to understand what might have happened during a crime can also be applied by, e.g., performing a virtual animation of the victim in the actual context, which can provide important information about possible dynamics during the event. Here, we present an overview of the different techniques and modalities used in forensic pathology in conjunction with crime scene investigations. Based on our experiences, the advantages and challenges of an image-based multi-modality approach will be discussed, including how their use may introduce new visualization modalities in court, e.g., virtual reality (VR) and 3D printing. Finally, considerations about future directions in research will be mentioned.

Estimating age at death by Hausdorff distance analyses of the fourth lumbar vertebral bodies using 3D postmortem CT images

The existing methods for determining adult age from human skeletons are mostly qualitative. However, a shift in quantifying age-related skeletal morphology on a quantitative scale is emerging. This study describes an intuitive variable extraction technique and quantifies skeletal morphology in continuous data to understand their aging pattern. A total of 200 postmortem CT images from the deceased aged 25-99 years (130 males, 70 females) who underwent forensic death investigations were used in the study. The 3D volume of the fourth lumbar vertebral body was segmented, smoothed, and post-processed using the open-source software ITK-SNAP and MeshLab, respectively. To measure the extent of 3D shape deformity due to aging, the Hausdorff distance (HD) analysis was performed. In our context, the maximum Hausdorff distance (maxHD) was chosen as a metric, which was subsequently studied for its correlation with age at death. A strong statistically significant positive correlation (P < 0.001) between maxHD and age at death was observed in both sexes (Spearman's rho = 0.742, male; Spearman's rho = 0.729, female). In simple linear regression analyses, the regression equations obtained yielded the standard error of estimates of 12.5 years and 13.1 years for males and females, respectively. Our study demonstrated that age-related vertebral morphology could be described using the HD method. Moreover, it encourages further studies with larger sample sizes and on other population backgrounds to validate the methodology.

Implementation of a personal identification system using alveolar bone images

OBJECTIVE

In recent years, personal identification has been performed using antemortem panoramic X-ray images and postmortem-CT images. Using these, we have developed a personal identification method that focuses on the alveolar bone. This study examined the effectiveness of this method and aimed to implement a reproducible system.

MATERIALS AND METHODS

For personal identification, a total of 633 CT images and panoramic X-ray images belonging to three groups with different conditions were used.　These images were 160 sets in the same person group and 96,820 in the other groups. The similarity of alveolar bone images was calculated using the landmark method of Procrustes analysis. The processes were system implemented and the methodology was validated.

RESULTS

The ability to identify between the same person group and other person groups showed 0.9769 as the area under the curve (AUC: ROC curve). At the cutoff value of 4.978, there was no false rejection rate, but false acceptance rate was slightly higher.

CONCLUSION

This method was useful as a screening method for personal identification. In addition, system implementation was efficient and reduced human error. In the future, we aim to realize a more efficient personal identification method using distortion-corrected images and including auto-detective landmarks using deep learning.

Performance and Robustness of Regional Image Segmentation Driven by Selected Evolutionary and Genetic Algorithms: Study on MR Articular Cartilage Images

The analysis and segmentation of articular cartilage magnetic resonance (MR) images belongs to one of the most commonly routine tasks in diagnostics of the musculoskeletal system of the knee area. Conventional regional segmentation methods, which are based either on the histogram partitioning (e.g., Otsu method) or clustering methods (e.g., K-means), have been frequently used for the task of regional segmentation. Such methods are well known as fast and well working in the environment, where cartilage image features are reliably recognizable. The well-known fact is that the performance of these methods is prone to the image noise and artefacts. In this context, regional segmentation strategies, driven by either genetic algorithms or selected evolutionary computing strategies, have the potential to overcome these traditional methods such as Otsu thresholding or K-means in the context of their performance. These optimization strategies consecutively generate a pyramid of a possible set of histogram thresholds, of which the quality is evaluated by using the fitness function based on Kapur's entropy maximization to find the most optimal combination of thresholds for articular cartilage segmentation. On the other hand, such optimization strategies are often computationally demanding, which is a limitation of using such methods for a stack of MR images. In this study, we publish a comprehensive analysis of the optimization methods based on fuzzy soft segmentation, driven by artificial bee colony (ABC), particle swarm optimization (PSO), Darwinian particle swarm optimization (DPSO), and a genetic algorithm for an optimal thresholding selection against the routine segmentations Otsu and K-means for analysis and the features extraction of articular cartilage from MR images. This study objectively analyzes the performance of the segmentation strategies upon variable noise with dynamic intensities to report a segmentation's robustness in various image conditions for a various number of segmentation classes (4, 7, and 10), cartilage features (area, perimeter, and skeleton) extraction preciseness against the routine segmentation strategies, and lastly the computing time, which represents an important factor of segmentation performance. We use the same settings on individual optimization strategies: 100 iterations and 50 population. This study suggests that the combination of fuzzy thresholding with an ABC algorithm gives the best performance in the comparison with other methods as from the view of the segmentation influence of additive dynamic noise influence, also for cartilage features extraction. On the other hand, using genetic algorithms for cartilage segmentation in some cases does not give a good performance. In most cases, the analyzed optimization strategies significantly overcome the routine segmentation methods except for the computing time, which is normally lower for the routine algorithms. We also publish statistical tests of significance, showing differences in the performance of individual optimization strategies against Otsu and K-means method. Lastly, as a part of this study, we publish a software environment, integrating all the methods from this study.

Automated Detection, Segmentation, and Classification of Pericardial Effusions on Chest CT Using a Deep Convolutional Neural Network

Pericardial effusions (PEFs) are often missed on Computed Tomography (CT), which particularly affects the outcome of patients presenting with hemodynamic compromise. An automatic PEF detection, segmentation, and classification tool would expedite and improve CT based PEF diagnosis; 258 CTs with (206 with simple PEF, 52 with hemopericardium) and without PEF (each 134 with contrast, 124 non-enhanced) were identified using the radiology report (01/2016−01/2021). PEF were manually 3D-segmented. A deep convolutional neural network (nnU-Net) was trained on 316 cases and separately tested on the remaining 200 and 22 external post-mortem CTs. Inter-reader variability was tested on 40 CTs. PEF classification utilized the median Hounsfield unit from each prediction. The sensitivity and specificity for PEF detection was 97% (95% CI 91.48−99.38%) and 100.00% (95% CI 96.38−100.00%) and 89.74% and 83.61% for diagnosing hemopericardium (AUC 0.944, 95% CI 0.904−0.984). Model performance (Dice coefficient: 0.75 ± 0.01) was non-inferior to inter-reader (0.69 ± 0.02) and was unaffected by contrast administration nor alternative chest pathology (p > 0.05). External dataset testing yielded similar results. Our model reliably detects, segments, and classifies PEF on CT in a complex dataset, potentially serving as an alert tool whilst enhancing report quality. The model and corresponding datasets are publicly available.