TECHNICAL NOTE ANTHROPOLOGY Romina Ciaffi, 1,2 B.Sc.; Danilo De Angelis, 1 Ph.D., D.D.S.; Pier Federico Gherardini, 3 B.Sc.; Giovanni Arcudi, 2 M.D.; Renato Nessi, 4 M.D.; Gian Paolo Cornalba, 4 M.D.; Marco Grandi, 1 M.D.; and Cristina Cattaneo, 1 Ph.D., M.D. Identification from Chest X-Rays: Reliability of Bone Density Patterns of the Humerus* ABSTRACT: A critical review of Kahana and Hiss’ study on identification from bone trabecular pattern and a test of their method conducted on the humerus are presented. Bone trabecular pattern was studied through the generation of a numerical file representing the gray scale. Using the corre- lation coefficient, several pairwise comparisons between numerical files were performed. The test gave nearly 30% of incorrect exclusions (the method did not recognize couples of radiographs belonging to the same subject) and 50% of misidentifications (the method recognized couples of radiographs belonging to different subjects, as belonging to the same subject); therefore, this research shows that at the present time, it is not possible to safely quantify identification through bone density patterns, of the proximal humerus taken from thoracic X-rays. Thus, an ‘‘easy’’—but dangerous—use of trabecular density patterns on this specific type of radiogram as an identification method should be currently avoided. KEYWORDS: forensic science, forensic anthropology, forensic radiology, human identification, trabecular bone pattern, quantification, thoracic radiograph, humerus In cases of criminal investigations that involve unidentified dece- dents, one of the main issues is the attribution of identity. When- ever the remains are fragmentary, burned, water damaged, or badly decomposed and visual ID by next of kin, fingerprint identification, or DNA analysis are not applicable, positive identification is usu- ally performed by forensic odontology. However, when this popular method cannot be applied, other methods of comparing antemortem and postmortem data are necessary. One of these is the comparison of bone morphology on antemortem and postmortem X-rays. In the case of bone morphology, however, it is impossible to give a quan- titative response, as in the case of DNA; neither are there clear guidelines concerning, for example, how many points of similarity are to be considered sufficient to claim identity. For this reason, several groups in different disciplines, including forensic anthropol- ogy and odontology, are testing the methods brought to trials (1–9). However, several authors claim that there is no need for quantifica- tion, because expert knowledge, training, and experience are the only reliable features for a correct identification (10), and visual comparison proves sufficient to achieve identification, assuming that the expert has a high level of training and experience in the forensic practice (8,11–13). Regardless, the quest for quantitative or semiquantitative methods still goes on. The aim of this technical note is to analyze and test the applicability of one of the few tech- niques in the field, proposed by Kahana and Hiss (14) and by Kah- ana et al. (15), which attempts to quantify bone morphological similarity, on more difficult yet more common radiographic mate- rial. Kahana and Hiss’ case reported in 1994 used hand radiographs and focused on metacarpal and phalanx bone architecture (14), while research conducted by Kahana et al., in 1998, used wrist radiographs and focused on the radius (15). Kahana et al. in their study state that the trabecular density pattern is unique to each indi- vidual and stable enough to be used as a forensic marker for the positive identification of human remains. They reached these con- clusions through the study of postmenopausal women wrist radio- graphs, taken annually, for a period of 6 years, with the aim of studying the bone mineral status of these women. The lowest corre- lation coefficient that Kahana et al. obtained comparing same sub- ject X-rays was 0.72 (r = 0.70–0.74, around 4% of same subject comparisons), while the highest correlation coefficient they obtained, comparing X-rays from different subjects, was 0.62 (r > 0.5, 5% of different subjects comparisons). Given these results, the authors claimed that the ‘‘cutting point’’ is 0.72 and the tech- nique provides a quantifiable criteria to be presented in a court of law. Furthermore observation of such patterns seems reliable as it has been demonstrated that bone mineralization density does not change in a period of 10 years (14,16). The method is therefore described as successful; however, wrist and hand X-rays are not easy to come by in cases of unidentified decedents. Thoracic X- rays on the other hand are much more common. Clearly, they are performed for the purpose of visualizing soft tissues and are not optimal for the study of skeletal structures. However, in many cases of unidentified decedents, antemortem X-rays that are most com- monly produced by family, physicians, or hospitals are thoracic 1 Labanof, Istituto di Medicina Legale, University of Milan, Via Mangia- galli 37, 20133 Milano, Italy. 2 Istituto di Medicina Legale, University of Rome ‘‘Tor Vergata’’, Via Montpellier 1, 00133 Rome, Italy. 3 Centre for Molecular Bioinformatics, Department of Biology, University of Rome ‘‘Tor Vergata,’’ Via della Ricerca Scientifica, 00133 Rome, Italy. 4 Institute of Radiological Sciences, University of Milan, and S. Paolo Hospital, via Di Rudini 8, Milan, Italy. *Presented at the XX th Congress of International Academy of Legal Medicine, August 23–26, 2006, in Budapest, Hungary. Received 4 July 2008; and in revised form 28 Feb. 2009; accepted 15 Mar. 2009. J Forensic Sci, March 2010, Vol. 55, No. 2 doi: 10.1111/j.1556-4029.2009.01297.x Available online at: interscience.wiley.com 478 Ó 2010 American Academy of Forensic Sciences