A Novel Method of Using Computed Tomography Data to Sort Commingled Skeletal Human Remains through Pair Matching

Abstract

In situations of commingled human skeletal remains, the first task of a forensic anthropologist is to accurately sort the remains and determine which bones belong to the same individual. Pair matching bilateral elements is one of the techniques often employed for this purpose. Current pair-matching methods are complicated by issues of subjectivity, bones of similarly sized individuals being incorrectly sorted, dimension reduction, need for highly trained practitioners, or the need for specific and expensive software applications. This research presents a novel and objective method of pair matching using 3D images rendered from computed tomography (CT) scans. Left and right bilateral elements were compared by mirroring left elements in Meshmixer and overlaying the mirrored left with a corresponding right element in CloudCompare. Next, the amount of discrepancy between the 3D images (meshes) was calculated in CloudCompare and summarized using two statistics, the mean of the distances between the vertices of one mesh to the closest vertices of the other (Dmean) and the standard deviation of those distances (Dsd). These statistics were used to build logistic regression models to be utilized in future predictive analysis to determine the probability of bones belonging to the same individual. The methods presented overcame some of the limitations of current pair matching methods mentioned while providing accurate, objective, and easily learned techniques to contribute to successful resolution of commingled human skeletal remains.