Assessing Inter‐ and Intra‐reliability in Frontal Sinus Outlines

Abstract

Human frontal sinuses are highly variable between individuals in terms of their size and shape. This idiosyncratic variation has been used in medicolegal fields for positive identification. Although qualitative comparisons are common practice, several 2D and 3D methods to objectively quantify their morphology have been proposed to statistically support such identifications. One approach has been to outline the superior border of the frontal sinus from traditional radiographs, using Elliptical Fourier analyses (EFA) of those frontal sinus outlines (FSOs) to characterize sinus shape. Several studies show this method quantitatively illustrates FSO uniqueness. To make such a method applicable to medicolegal cases, however, inter‐ and intra‐observer reliability in the tracing of FSOs must first continue to be evaluated.This study uses digital radiographs of 40 crania in Caldwell position from Des Moines University’s anatomical collection to assess inter‐ and intra‐observer variations in FSO tracings. Five crania had bilateral aplasia; two were too difficult to see. FSOs were traced in the remaining 33 crania by two researchers using ImageJ, with the superior orbital border demarcating the inferior boundary. FSOs were imported into SHAPE‐software, and underwent EFA (20 harmonics) and principal component analyses (PCA). PC1 (50.34% of the variance) reflected height‐vs‐breadth relationships, while PC2 (13.40%) reflected differences in outline complexity between researchers; ANOVAs confirmed these significant differences along PC2 (F=10.879; p<0.0001). Further investigation revealed that one observer tended to follow the lobe contours more detailed, tracing further into the septa between lobes. With this knowledge, a second set of tracings by the same researchers, both better accounting for complexity, were undertaken to assess whether more specific instructions can control for such deviations. Subsequent PCAs and ANOVAs indicated no significant differences in PC1 or PC2 within or between individual tracings (p‐values > 0.05). However, several FSO‐tracings still plotted relatively far apart along the PC1 and PC2 axes. These “problematic” sinuses were often smaller in size or had smaller lobes. Interestingly, the issue lies primarily with inferior border placement— not the sinus itself. In such cases, small lobes near the inferior boundary were not always included in the FSO‐tracing, greatly altering FS size and shape as captured by EFA.This information helps support the use of FSOs in forensic identification, but only when appropriate considerations are made. Similar to previous studies, these results indicate that investigators should be aware that FSOs of smaller sinuses may not be as reliable; this study also indicates large sinuses with smaller lobes may also be problematic. Intra‐ and inter‐observer reliability of the placement of the superior orbital border and potential errors beyond tracing deviations need further investigation to fully understand potential errors and any major implications in the use of FSO in medicolegal identifications, given the high stakes of misidentifying an individual or erroneously excluding a true identity.Support or Funding InformationFunding support provided by IOER‐Grant 03‐19‐01.