The Effects of Cranial Orientation on Forensic Frontal Sinus Identification as Assessed by Outline Analyses

Personal identification using frontal sinus coding methods: The effect of mixed image modality comparisons

Several code-based methods have been created for comparing the frontal sinus in skeletal identification scenarios. However, little is known regarding matched-pair accuracy rates of these methods or how varying image modalities may affect these rates. The goals of this study were to validate the exclusion rates and to establish matched-pair accuracy rates of two well-cited coding methods, Cameriere et al. [23] and Tatlisumak et al. [24]. Additionally, individual variables were assessed for consistency in scoring between image modalities. Using a sample of U.S. African American, Native American, and European American females and males (n = 225), we examined individual variable scoring and string codes between two different image modalities (radiographs and CT-based 3D models). Arcades showed poor scoring consistency between modalities (p < 0.001). Although exclusion rates were similar to those reported in the original studies (93%-96%), matched-pair accuracy rates were low (13%-18%). None of the demographics (collection, sex, age, ancestry, and orientation) had an effect on the odds of a match. Interobserver and intraobserver analyses showed moderate to near-perfect agreement for all variables except supraorbital cells, which had minimal to no agreement. Currently, we do not recommend the application of these frontal sinus coding methods independent of other supporting identification methods given low variable consistency and accuracy rates. Visual identification should still be used to include or exclude an identification when using the frontal sinus.

Variation, sexual dimorphism, and enlargement of the frontal sinus with age in adult South Africans

OBJECTIVES

To document frontal sinus volume (FSV) in a sample of sub-Saharan Africans with a view to evaluating claims that such populations exhibit comparatively small sinuses. This study also addresses questions related to sexual dimorphism, incidence of sinus aplasia, and the possibility that FSV continues to increase through adulthood.

MATERIALS AND METHODS

FSV was measured from CT scans of adult crania from the Dart Collection. Sex and age were known for each individual. Linear cranial dimensions were used to compute a geometric mean from which a scaled FSV was computed for each cranium.

RESULTS

FSV does not differ significantly between sexes, but females exhibit a higher incidence of aplasia. There is considerable variation in FSV in this sample, with the average ranking among the higher means reported for other population samples. The incidence of FS aplasia falls within the range of values recorded for other population samples. Although our study is cross-sectional rather than longitudinal, there is strong evidence that FSV continues to increase with age throughout adulthood.

DISCUSSION

The FSV mean of our sample contradicts the notion that sub-Saharan Africans possess small sinuses. In a global context, geography (climate and altitude) does not appear to be related to FSV. The absence of sexual dimorphism in our sample is unexpected, as significant dimorphism has been reported for most other population samples. Our results support other indications that the frontal sinus continues to expand throughout adulthood, especially in females, and that it is likely due to bone resorption.

Evaluating artificial intelligence for comparative radiography

INTRODUCTION

Comparative radiography is a forensic identification and shortlisting technique based on the comparison of skeletal structures in ante-mortem and post-mortem images. The images (e.g., 2D radiographs or 3D computed tomographies) are manually superimposed and visually compared by a forensic practitioner. It requires a significant amount of time per comparison, limiting its utility in large comparison scenarios.

METHODS

We propose and validate a novel framework for automating the shortlisting of candidates using artificial intelligence. It is composed of (1) a segmentation method to delimit skeletal structures' silhouettes in radiographs, (2) a superposition method to generate the best simulated "radiographs" from 3D images according to the segmented radiographs, and (3) a decision-making method for shortlisting all candidates ranked according to a similarity metric.

MATERIAL

The dataset is composed of 180 computed tomographies and 180 radiographs where the frontal sinuses are visible. Frontal sinuses are the skeletal structure analyzed due to their high individualization capability.

RESULTS

Firstly, we validate two deep learning-based techniques for segmenting the frontal sinuses in radiographs, obtaining high-quality results. Secondly, we study the framework's shortlisting capability using both automatic segmentations and superimpositions. The obtained superimpositions, based only on the superimposition metric, allowed us to filter out 40% of the possible candidates in a completely automatic manner. Thirdly, we perform a reliability study by comparing 180 radiographs against 180 computed tomographies using manual segmentations. The results allowed us to filter out 73% of the possible candidates. Furthermore, the results are robust to inter- and intra-expert-related errors.

Frontal Sinus Morphological and Dimensional Variation as Seen on Computed Tomography Scans

Frontal sinus variation has been used in forensic anthropology to aid in positive identification since the 1920s. As imaging technology has evolved, so has the quality and quantity of data that practitioners can collect. This study examined frontal sinus morphological and dimensional variation on computed tomography (CT) scans in 325 individuals for assigned sex females and males from African-, Asian-, European-, and Latin American-derived groups. Full coronal sinus outlines from medically derived CT images were transferred into SHAPE v1.3 for elliptical Fourier analysis (EFA). The dimensional data were measured directly from the images using the MicroDicom viewer. Statistical analyses—Pearson’s chi-square, ANOVA, and Tukey post hoc tests—were run in R Studio. Results indicated that 3.7% lacked a frontal sinus and 12.0% had a unilateral sinus, usually on the left (74.3%). Additionally, no statistically significant morphological clustering using EFA was found based on assigned sex and/or population affinity. However, there were statistically significant differences dimensionally (height and depth) when tested against assigned sex and population affinity, indicating that the interactive effects of sexual dimorphism and adaptive population histories influence the dimensions but not the shape of the frontal sinus.

Technical Modifications for the Application of the Total Difference Method for Frontal Sinus Comparison

Despite being used in personal identification since the 1920s, frontal sinus-based methods are rarely validated. This study is a validation test of the Total Difference Method (TDM). The posterior-anterior radiographs of 10 adults were assessed by two observers using three modes: the Freehand Mode largely followed the original protocols; the Overlay Mode utilized a tracing overlay; and the Semi-Auto Mode used the overlay and macro, walking the user through multiple steps. The modes were evaluated for the time taken to complete each image and the accuracy and repeatability of the line lengths, midline assessment, and angle placement. The repeated measures analysis of variance results for the intra-observer error revealed differences in bias in the angle placement and line length mean error between the rounds and modes. The differences between the rounds were approximately consistent for each mode, suggesting observer error. Significant differences in the inaccuracy of the angle placement and the line lengths between observers by mode were evident; in post hoc testing, the Freehand Mode and Overlay Mode had the greatest error in both variables (p.adj < 0.0001). The Semi-Auto Mode retained no significant error for angle inaccuracy and had the fewest errors for line length inaccuracy (p.adj < 0.01). When using the Semi-Auto Mode, the time was 46.1% faster than that of the Freehand Mode and 34% faster than that of the Overlay Mode (F2,18 = 52.71, p < 0.0001); time slightly improved with method familiarity. The results suggest that the technique required for the TDM can reliably be repeated, and the semi-automated macro improves accuracy and efficiency, but only after the users are familiar with the method and macro procedures. All resources needed to repeat this study are freely available on GitHub.