Master of Science (MS)
First Committee Member
Second Committee Member
Third Committee Member
Fourth Committee Member
Number of Pages
Introduction:Estimation of facial soft tissue appearance from human skeletal remains is often necessary in forensic identification. This process has been referred to as facial reconstruction or facial approximation and is a branch of forensic facial anthropology. Original methods for facial approximation originated in nineteenth century Europe and consisted of artists shaping clay over skull models using average soft tissue depths measured in cadavers. The last two decades have introduced numerous computerized techniques that have digitized this process while attempting to accurately and objectively define the relationship between a skull and its overlying soft tissue. This pilot study describes a method of facial approximation that combines cephalometric techniques for characterization of the craniofacial complex commonly used in the field of orthodontics with a database of cone-beam computed tomography (CBCT) skull images. Facial likenesses for an unknown skull are automatically located within the database by comparing cephalometric values recorded on the unknown skull with those within the database. A recently proposed method of sex determination based on the anatomy of the mastoid process, glabellar process, and frontal sinus area is also applied to the sample used in this study.
Methods:A database consisting of one-hundred (100) cone-beam computed tomography (CBCT) skull images of Hispanic female patients of the University of Las Vegas, Nevada School of Dental Medicine Orthodontic Department [age range 8 to 23 years (mean 13.5 years)] was constructed. A cephalometric analysis consisting of twelve (12) landmarks and nineteen (19) skull measurements [sixteen (16) angular and three (3) proportional] was defined and applied to all database entries. Facial approximations were created for three skulls by sequentially removing three (3) random entries from the database and treating these as unknown (leave-one-out cross validation). A weighted least-sum-of-squares (WLSS) regression algorithm was applied to measure the cephalometric similarity between each entry in the database and the unknown skull data to find the three (3) most cephalometrically similar skulls in the database (three closest matches). Accuracy was assessed through expert face pool resemblance ranking. Soft tissue profiles associated with the three best matches were grouped with three random database entries to create a face pool array of size six (6) for each unknown. Fourteen (14) post-doctoral orthodontic graduate students were utilized as expert face pool evaluators. Sex determination accuracy was then assessed by comparing the values of eight (8) cephalometric measurements taken on this sample to those already described and proven efficacious on other samples in the literature.
Results:Intraexaminer reliability was acceptable for all cephalometric measurements. Expert face pool resemblance rankings results implied that the described process was able to select database entries that approximated the unknown face better than random database entries. In Face Pools One, Two, and Three the three highest ranked faces contained two, two, and three algorithm-selected faces, respectively. Sex determination data recorded on this sample was comparable to data described in the literature.
Conclusions:Contemporary methods of facial approximation have shown that estimation of soft tissues from skeletal data can be achieved by employing computationally and graphically complex techniques. It now also seems plausible to rapidly estimate the general shape of an unidentified skull's facial profile by comparison of the unknown skull's cephalometric data to those in a database of orthodontic patients. Further research involving the described method is warranted.
Cephalometry; Facial approximation; Facial reconstruction (Anthropology); Forensic anthropology; Forensics; Identification; Orthodontics
Anatomy | Anthropology | Criminology and Criminal Justice
Nejat, Behzad, "A Forensic Identification Utility to Create Facial Approximations using Cone-Beam Computed Tomography of 100 Hispanic Females: A Pilot Study" (2012). UNLV Theses, Dissertations, Professional Papers, and Capstones. 1763.