As submitted by the proposer:
Statement of the Problem: For many cases of skeletonized remains, all efforts of identification have been unsuccessful, and facial approximation offers a chance that someone may recognize the decedent. A significant problem with facial approximation is the lack of scientific guidelines and standardized protocols. More importantly, there is a lack of comprehensive, large-scale studies of craniofacial variation of modern Americans. Most studies have been limited by small sample sizes, analyses of one facial feature, or samples from outside of the U.S. Therefore, we propose a comprehensive investigation of craniofacial variation in modern Americans by examining a large collection of head CT scans. The primary goal of this project is to find bone measurements and features that more reliably predict soft tissue features.
Subjects: Head CT scans from The Cancer Imaging Archive will be used for this study. The database consists of about 280 anonymous male and female subjects of known age and sex.
Although ancestry is not available, our study proposes a morphologically-driven approach that will focus on feature variation and preclude the need for ancestry information.
Partnerships: For this project, we have enlisted a forensic artist to evaluate non-metric traits of the bone and skin. This collaboration will help guide the collection of metric data and ensure that measurements are practical for forensic artists.
Research Design and Methods: Phase 1 (1st month) will involve preparation of the CT scans by editing, evaluating scan quality, and generating 3D bone and skin models. Phase 2 will consist of data collection using Mimics software through placement of anthropometric landmarks on 3D bone and skin models, and collection of interlandmark distances (ILDs) and associated angles.
Standard and novel landmarks will be used to evaluate bone ILDs commonly used to predict skin
ILDs and explore new relationships. We will also include non-metric evaluations of facial variation to be carried out primarily by the forensic artist.
Analysis: Analyses will consist of descriptive statistics, correlation analysis, and multiple linear regressions. We will use multiple linear regression to find sets of bone ILDs that are the best predictors of skin ILDs of interest.
Products, Reports, and Data Archiving: Manuscripts will be prepared for publication in peer-reviewed journals in Year 2, and results will be presented at AAFS and IAI. Another goal is to produce a visual reference organized by facial feature and supplemented with metric data where applicable. Archived data will consist of 3D landmark coordinates.