Biology Reference
In-Depth Information
dentition must contain enough information to identify the suspect uniquely; and (2) the
information in the dentition must be faithfully recorded in the bitemark to allow successful
use of it as evidence.
In the effective post-mortem identifications, the forensic dentist often has access to
detailed x-rays of the entire dentition, as well as information about the location and nature
of dental work including fillings, root canals and bridges. This information is generally
accepted as unique and adequate to identify a specific individual, having a long and non-
controversial history of success. In contrast, in bitemark analysis, typically all that is avail-
able are impressions left by the incisal (cutting) surfaces of the twelve most anterior teeth in
the human dentition (i.e., the six most forward teeth of the upper and lower jaws). Thus, a
bitemark impression typically contains far less information than is available from all surfaces
and internal structures of the whole dentition (normally 28
wisdom teeth).
The study most often cited to support the claim of uniqueness of the portion of the
human dentition that creates bitemarks examined bitemarks in wax produced by 384 den-
titions (
Rawson et al., 1984
). In this study, Rawson et al
.
superimposed or registered all
the measured dentitions in a standard orientation, and then determined the
x
and
y
coor-
dinates of the midpoint of each tooth (to
32 teeth,
6
1mm) and the relative angular orientation of
6
each tooth within
2.5 degrees (see
Rawson et al., 1984; or Bush et al., 2011b
, for complete
details). Rawson et al. used a form of baseline registration that was fairly sophisticated for
1984, but not consistent with methods developed since then. From this information, they
calculated the number of possible states (or positions) each tooth position and angle could
occupy by the range of observed values divided by the measurement resolution. From the
very large calculated number of possible states, Rawson and colleagues concluded that the
human anterior dentition was effectively unique to individuals.
In reconsidering this work,
Bush et al. (2011b)
noted that the Rawson et al. model did
not include correlation among teeth and assumed uniform distribution models. Bush and
colleagues also noted that no attempt was made to search through the data to see if any
dentitions matched. When
Bush et al. (2011b)
searched their collected data, they found
matches between lower dentitions in relatively small data sets. Later, using two- and
three-dimensional geometric morphometric data, they looked for matches in collections of
measured human dentitions (
Bush et al., 2011a,b,c; Sheets et al., 2011, 2013
). In a later
study (
Sheets et al., 2013
), a collection of 1106 paired sets of three-dimensional scanned
maxillary and mandibular dentitions (upper and lower, respectively) were obtained from
a commercial dental laboratory, which had used these scans to produce occlusal guards
(night guards). The dental models were three-dimensional scans from private practice
patients across the USA. (All necessary Human Subject Institutional Review Board proto-
cols were completed for this project and exemption was granted and all patient identifying
information was stripped from the data). This was a sample of convenience that contained
a wide range of alignment patterns, from relatively straight to fairly mal-aligned. After ini-
tial matching studies, seven of these specimens were identified as being repeated scans of
the same individuals and removed from the study, leaving 1099 distinct individuals.
Landmarks were recorded on the three-dimensional scanned dentitions by placing 10
data points along the incisal edge of each of the six anterior teeth, using the digitizing
program Landmark (Institute for Data Analysis and Visualization, UC
Davis, 2011
,
Figure 14.1
). The dentitions were rotated in three-dimensional space within the software
6
