Sex differences in proximal humeral outline shape: Elliptical Fourier functions

Citation
H. Tanaka et al., Sex differences in proximal humeral outline shape: Elliptical Fourier functions, J FOREN SCI, 45(2), 2000, pp. 292-302
Citations number
34
Categorie Soggetti
Research/Laboratory Medicine & Medical Tecnology
Journal title
JOURNAL OF FORENSIC SCIENCES
ISSN journal
00221198 → ACNP
Volume
45
Issue
2
Year of publication
2000
Pages
292 - 302
Database
ISI
SICI code
0022-1198(200003)45:2<292:SDIPHO>2.0.ZU;2-4
Abstract
A method is presented for the numerical analysis of sex differences in size and shape of the proximal humeral outlines using elliptical Fourier functi ons (EFFs). A skeletal sample consisting of right and left humeri pairs of 69 individuals, 36 males and 33 females, was used. The proximal superior vi ew in the plane of the proximo-distal axis of each humerus was photographed and then 54 boundary points were located on the two-dimensional outline tr acings. These points were digitized and used to compute EFFs with 27 harmon ics. From the EFFs, a set of expected points on the proximal humeral outlin e was generated using the centroid as an origin. Superimposition of the mal e and female outlines on this centroid provided a detailed picture of the r elative sex differences in size and shape with respect to that center. The bounded area of the proximal humeral outline showed statistically signi ficant sex differences. Additionally, statistical results of the amplitudes derived from the "area-standardized" EFFs and visual assessments of the me an outline plots indicated significant sex differences in shape of the prox imal humeral outlines. Focusing on localized regional differences. the grea ter tubercle was located more postero-medially and the lesser tubercle was located more anteriorly in the males compared to the females. Sex determina tions from the proximal humeri were also examined with discriminant functio ns based on the amplitudes, which represent shape characteristics of the ou tline, and the bounded area. Using a cross-validation method, predictions o f the percentages of cases correctly classified with the discriminant funct ions were ranged from 92.8% to 95.7% for the right and left humeral data. T hese results suggest that differences in size and shape of the proximal hum eral outlines may be better predictors of sex when compared with convention al measurements of the humerus.