ABSTRACT
This study presents a technique developed for 3-D imaging and quantitative comparison of human dentitions and simulated bite marks. A sample of 42 study models and the corresponding bites, made by the same subjects in acrylic dental wax, were digitised by laser scanning. This technique allows image comparison of a 3-D dentition with a 3-D bite mark, eliminating distortion due to perspective as experienced in conventional photography. Cartesian co-ordinates of a series of landmarks were used to describe the dentitions and bite marks, and a matrix was created to compare all possible combinations of matches and non-matches using cross-validation techniques. An algorithm, which estimated the probability of a dentition matching its corresponding bite mark, was developed. A receiver operating characteristic graph illustrated the relationship between values for specificity and sensitivity. This graph also showed for this sample that 15% of non-matches could not be distinguished from the true match, translating to a 15% probability of falsely convicting an innocent person.
