Evaluating the Usefulness of CT Data for Trauma Analysis and Sex Estimation
BORIS DOI
Abstract
Anthropological research constantly attempts to refine methods for the estimation of an individual's biological profile: their age-at-death, sex, stature and ancestry. Traditionally, protocols for the estimation of the biological profile have been developed based on identified osteological collections. However, many of these samples represent non-modern populations, which may curtail the validity of research outcomes, especially when the focus is forensic. A valid alternative to avoid this bias is the use of virtual osteological collections. They are usually based on tomographic data, surface scans or photogrammetric images and allow for easier data sharing, perpetual access and non-invasive analysis. However, the increasing use of imaging techniques in anthropological research raises some questions concerning their usefulness, calling for a test quantifying the errors associated with the use of these types of datasets. Here, we explore this topic starting from the following two research questions: a) Can forensic anthropologists without a radiological background detect skeletal lesions caused by sharp force trauma on postmortem computed tomography (PMCT) data from an identified forensic dataset?; b) Are observations made on virtual models interchangeable with those collected on the actual skeletal element for the estimation of sex? To address the first question, two observers analyzed a retrospectively collected sample (n=41) on two-dimensional (2D) slices and three-dimensional (3D) reconstructions of PMCT data. We used the forensic autopsy and radiology reports containing information about the skeletal sharp force lesions as a baseline for the comparison with our observations. For the second question, we applied seven different metric and morphoscopic sex estimation methods to 200 pelves and 223 skulls of archaeological origin and comparedthe results with those based on the CT based virtual models of the same specimens. For comparative purposes, we included models obtained from surface scans of 39 pelves and 50 skulls selected from the same sample. Of the 71 skeletal sharp force lesions described in the forensic reports of the 41 sharp force
cases, we detected between 30 (42%) and 41 (58%), depending on the viewing modality, of which the 2D slices proved best for the purpose. The intra- and interobserver agreements are high (κ 0.474-1). For the estimation of sex, we found the highest intermodality agreement for the metric method on the pelvis (rTEM <5%). Morphoscopic traits were more prone to errors, especially on the skull (κ 0.479-0.594). Moreover, the two virtual modalities (CT and 3D surface scans) yielded a higher agreement (κ 0.756) than either of them compared to dry bone. The relatively low detection rate of sharp force bone lesion implies that pertinent training is crucial if PMCT data are used on a regular basis. At the same time, we found differences between the two forensic reports in the number of detected skeletal bone lesions similar to our observations. For the sex estimation study, we conclude that CT and 3D surface scans are highly interchangeable because of the visual-only observational approach on both modalities.
cases, we detected between 30 (42%) and 41 (58%), depending on the viewing modality, of which the 2D slices proved best for the purpose. The intra- and interobserver agreements are high (κ 0.474-1). For the estimation of sex, we found the highest intermodality agreement for the metric method on the pelvis (rTEM <5%). Morphoscopic traits were more prone to errors, especially on the skull (κ 0.479-0.594). Moreover, the two virtual modalities (CT and 3D surface scans) yielded a higher agreement (κ 0.756) than either of them compared to dry bone. The relatively low detection rate of sharp force bone lesion implies that pertinent training is crucial if PMCT data are used on a regular basis. At the same time, we found differences between the two forensic reports in the number of detected skeletal bone lesions similar to our observations. For the sex estimation study, we conclude that CT and 3D surface scans are highly interchangeable because of the visual-only observational approach on both modalities.
Date of Publication
2024
Theses Type
dissertation
Language(s)
en
Author(s)
Faculty/Graduate School
Access(Rights)
restricted
Primary OA Publication
false