LINEAR DENSITOMETRY AND DIGITAL IMAGE-PROCESSING OF PROXIMAL FEMUR RADIOGRAPHS - IMPLICATIONS FOR ARCHAEOLOGICAL AND FORENSIC ANTHROPOLOGY

Age- and sex-related patterns of proximal femur trabecular bone loss h ave previously been used to establish radiographic reference standards for estimating age at death for human skeletal remains. Such standard s are of interest to both anthropologists and forensic scientists. How ever, osteopenia as a physiological phenomenon is dependent on numerou s genetic, environmental, and cultural factors. Thus, while general ag e-and sex-related trends can be clearly observed for trabecular bone l oss, such patterns also demonstrate marked variation among individuals of both sexes at all ages. Moreover, clinical evidence shows that rat es of bone loss are not steady but episodic, and that radiographically ''normal'' (i.e., young adult) patterns of trabecular bone architectu re can also exist in femora of older individuals, particularly within samples of African origin. In this study, adult proximal femur radiogr aphs were used to explore patterns of age- and sex-related proximal fe mur cancellous bone involution among a sample of 66 African-American i ndividuals from the Terry collection (33 males and 33 females), rangin g in age from 19 to 71 years. The proximal femur radiographs of these subjects were analyzed by digital image processing (DIP), and the resu lts were compared to those obtained by laser linear densitometric anal yses (LDA) previously performed on the same series (Macchiarelli et al ., 1987). Results of LDA and DIP analyses indicate (a) more pronounced bone density decrease in females; (b) sex- and site-specific structur al patterns of proximal femur trabecular bone loss; (c) a high level o f individual variability, in which predicted age deviated from real ag e by as much as 22.26 (males) and 30.78 years (females); (d) a moderat e linear correlation with age for all the variables analyzed; and (e) an average discrepancy between known age and predicted age (measured b y root mean squared residual values) of 10.34 (males) and 12.71 years females) for the most satisfactory DIP analysis parameter (recorded at the center of the femoral neck in the equalized images). With regard to estimating adult age at death from radiographic images of proximal femora, it is argued that the amount of useful age information reporte d for this criterion has been overestimated. (C) Wiley-Liss, Inc.