Efficacy of automated three-dimensional image reconstruction of the femur from postmortem computed tomography data in morphometry for victim identification

Abstract

Besides conventional radiology, postmortem computed tomography (PM-CT) is nowadays widely used for victim identification in forensic routines to detect anatomical characteristics and specific pathologies as well as to estimate the stature and sex. A major advantage of PM-CT is virtual reconstruction of skeletal structures independent of the status of recovered remains. The present study investigated the efficacy of a recently provided automated analyzer to reconstruct three-dimensional (3-D) images using CT data for skeletal morphometry, measuring the whole bone mass volume of the femur (60 bones in 30 cases) as an index of reproducibility. Manual cursoring could reconstruct bilateral femurs with high reproducibility, showing mass volume fluctuations by repetition and between two independent observers of 0.2-2.1% and 3.5-6.7%, respectively, partly depending on the data analysis system, but was time-consuming, while automated reconstruction was very rapid and highly reproducible virtually without detectable fluctuation; there was a high correlation between bone mass volumes reconstructed by manual and automated procedures (r=0.9976, p<0.0001). The reproducibility of the automated procedure was 98.64-100.81% in 5 cases scanned twice under the same CT conditions. Preliminary analysis showed a substantial correlation of the whole femur mass volume with the body height and a significant sex-related difference in the femur mass volume/body height ratio (males>females). These findings indicate the accuracy and practical feasibility of the automated procedure to reconstruct single bone 3-D CT images for virtual skeletal morphometry in victim identification.