The accuracy and repeatability of lodox statscan osteometric measurements

Abstract

The rise of the use of radiographic imaging techniques in forensic science, propelled forensic anthropologists to explore virtual imaging modalities, in order to estimate biological profiles. This includes low-dosage X-ray (Lodox), which is currently operational in South African medico-legal laboratories. This research aimed to compare the repeatability and accuracy of osteometric measurements taken from Lodox images of macerated bones, with that of the corresponding dry bone material.A total of 50 full-body Lodox images with corresponding dry bones were selected from the Kirsten Skeletal Collection. Forty-six standard osteometric measurements were collected from the Lodox images and the corresponding dry bone. The sample was equally distributed by sex and consisted of adults, aged 23-49. Where applicable, the left and right skeletal elements were both collected, resulting in 84 measurements collected from three datasets; the skull (n=9), pectoral girdle and upper limb (n=36), as well as the pelvic girdle and lower limb (n=39). Using technical error of measurement (TEM), Lodox measurements were deemed repeatable (intra-observer rTEM < 1.5% and inter-observer rTEM < 2.0%), with exceptions of the intra- and inter-observer results of the; maximum ramus height, minimum ramus breadth, maximum length of clavicle, physiological length of ulna, maximum vertical diameter of the head of humerus, maximum diameter of the femoral head and the maximum and minimum diameter of the clavicle, humerus, radius, ulna, femur and fibula. Using percentage differences, 72.6% of the measurements collected had percentage differences less than 10%. This consisted of the long bone measurements along the scan direction/y-axis. No significantdifference (p>0.05) was noted with 25% of the measurements collected. Most of these measurements were small midshaft measurements collected along the

x-axis. Majority of the measurements were deemed repeatable; however, these measurements were not accurate, likely due to the use of 2D images