Facial Interlandmark distances on CT soft tissue models of 388 living persons
Description
Averaged facial ILDs on CT-derived models and on ReFace approximations were reported by Simmons-Ehrhardt et al. [13] for 388 living subjects in the U.S.A. These ILDs (Fig. 1) were defined by 12 standard anthropometric landmarks [22]. The CT database comprised both sexes and four ancestry groups, African female (n=50), African male (n=48), Asian female (n=48), Asian male (n=47), European female (n=49), European male (n=48), Hispanic female (n=49), and Hispanic male (n=49). Each ReFace facial approximation was determined within the appropriate demographically related dataset, but from which data for each target was temporarily removed. The approximations were compared with the known CT-derived 3D models, testing the verisimilitude of ReFace approximations by modeling a facial recognition system. [13] T.L. Simmons-Ehrhardt, T. Flint, C.P. Saunders, K.L. Monson, Quantitative accuracy and 3D biometric matching of 388 statistically estimated facial approximations of live subjects, Forensic Img 21 (2020) 200377. doi:10.1016/j.fri.2020.200377. [22] L.G. Farkas, Anthropometry of the Head and Face, 2nd ed., Raven Press, New York, 1994.
Files
Steps to reproduce
Adult CT scans were collected between 2003-2009 by GE Global Research for use as a reference database for ReFace. Collection and intended use of these anonymized data were approved by the institutional research boards of the partner institutions, and each subject signed an informed consent agreement. The CT scans were acquired from multiple institutions and were collected under a variety of scanning protocols, with slice thicknesses ranging from 0.98 mm to 6.00 mm, slice increments ranging from 0.10 mm to 5.00 mm, pixel size ranging from 0.449 mm to 0.586 mm, and three X-Y image resolutions [5]. Mimics v.11.1 and v.12.0 software (Materialise, Ann Arbor, MI) was used to convert CT scans to separate 3D bone and skin models. Bone was segmented from soft tissue using a threshold value of 226 Hounsfield units. Segmentation masks were edited prior to 3D reconstruction to remove large dental artifacts and vertebrae to allow better access to craniometric landmarks. The simulation module of Mimics was used to place landmarks on the digital 3D skulls [12]. [5] C.L. Parks, A.H. Richard, K.L. Monson, Preliminary assessment of facial soft tissue thickness utilizing three-dimensional computed tomography models of living individuals, Forensic Sci. Int. 237 (2014) 146.e1-146.e10. doi:10.1016/j.forsciint.2013.12.043. [12] T. Simmons-Ehrhardt, C. Ehrhardt, K. Monson, Evaluation of the suitability of cranial measurements obtained from surface-rendered CT scans of living people for estimating sex and ancestry, J Forensic Radiol Imag 19 (2019) 100338. doi:10.1016/j.jofri.2019.100338.