Correlations of facial soft tissue thicknesses with craniometric dimensions improve craniofacial identification estimates: Fact or fiction?

Abstract

Linear regression (LR) models that use cranial dimensions to estimate facial soft tissue thicknesses (FSTTs) have been posited by Simpson and Henneberg to assist craniofacial identification. For these regression equations to work well, the independent (craniometrics) and dependent (FSTTs) variables must be tightly correlated; however, such relationships have not been routinely demonstrated for adult humans. To examine the strength of these relationships further, this study employed magnetic resonance (MR) imaging to unambiguously measure cranial dimensions and FSTTs for 38 adult cadavers. This contrasts with prior published use of (a) spreading calipers to compress the face in an attempt to measure cranial dimensions through the soft tissues of the head and (b) needle puncture to measure the FSTT (a + b = legacy methods). To provide direct comparisons to prior work, this study also conducted legacy measurements for reproducibility tests. Previously published LR models were not supported by either the legacy or MR data. In the MR data, correlations between the cranial dimensions and FSTTs were generally weak with a mean absolute r = 0.13 and r range = -0.40 to 0.33. Compared to previously published arithmetic means, LR models did not improve FSTT estimates in the MR sample (SEE for LR = 5.5 mm; SEE for arithmetic mean = 5.2 mm). Using the latest MR imaging technologies, these results underscore prior warnings that cranial dimensions hold weak correlations with FSTTs and possess limited utility for improved FSTT estimation over and/or above arithmetic means.