An Empirical Examination of Frontal Sinus Outline Variability Using Elliptic Fourier Analysis: Implications for Identification, Standardization, and Legal Admissibility

The comparison of frontal sinus radiographs for positive identification has become an increasingly applied and accepted technique among forensic anthropologists, radiologists, and pathologists. However, the current method of outline comparison by visual assessment fails to meet evidence admissibility guidelines as set forth in the 1993 case of Daubert v. Merrell-Dow Pharmaceuticals, Inc. Specifically, no empirical testing of the uniqueness of frontal sinus outlines has ever been performed, there has been no evaluation of the probability of misidentification using the technique, there are no standards controlling the technique’s operation, and there are no subjective standards for confirming or rejecting a putative identification. Despite the fact that identifications based upon frontal sinus radiograph comparisons have been routinely accepted by scientists, medical examiners and law enforcement officers, these shortcomings could pose serious problems if forensic scientists were ever called upon to testify regarding such an identification in trial.

This study investigated frontal sinus outline variability using Elliptic Fourier Analysis (EFA), a geometric morphometric approach that fits a closed curve to an ordered set of data points, generating a set of coefficients that can be treated as shape descriptors used as variables in discriminatory or other multivariate analyses, or used to reproduce the outline. By modeling 2-dimensional representations of frontal sinuses (as seen in posterior-anterior cranial radiographs) as closed contours by digitizing their outer borders, differences in their shapes were assessed quantitatively by comparing the Euclidean distances between the EFA-generated outlines. The probability of misidentification was assessed using likelihood ratios and posterior probabilities based on the EFA coefficients.

Results showed that there is a quantifiable and significant difference between the shapes of different individuals’ frontal sinus outlines as represented by Euclidean distances, since distances between outlines of different individuals were shown to be significantly larger than those between replicates (simulated antemortem and postmortem) of the same individual. Likelihood ratios using EFA coefficients showed that the probability of a frontal sinus match given the correct identification versus the probability of a match from the population at large was very high, and therefore the probability of misidentification was very low.

This study concluded that for individuals with sufficiently remarkable frontal sinus outlines, using EFA coefficients of digitized frontal sinus outlines to estimate the probability of a correct identification, and thereby confirm or reject a presumptive identification, is a reliable technique. Given these results, EFA comparison of frontal sinus outlines is recommended when it may be necessary to provide quantitative substantiation for a forensic identification based on these structures.