The use of craniofacial superimposition for disaster victim identification

Highlights

• Comparison of manual and computerised craniofacial superimposition for DVI.

• Twenty skulls compared to face pools.

• Profile and three-quarter views were the most valuable facial images.

• Poor identification rate suggests MPUB database images would not be optimal.

• Multiple ante-mortem images with a variety of facial expression should be utilised.

Abstract

Skull-to-face comparison is utilised for human identification where there is a suspected identity and the usual methods of identification, such as DNA or dental comparison, are not possible or practical. This research aimed to compare the reliability of manual and computerised craniofacial superimposition techniques and to establish the application of these techniques for disaster victim identification, where there may be a large database of passport-style images, such as the MPUB Interpol database. Twenty skulls (10 females; 10 males) were utilised from the William Bass Skeletal Collection at the University of Tennessee and compared to face pools of 20 face photographs of similar sex, age and ethnic group. A traditional manual photographic method and a new 3D computer-based method were used. The results suggested that profile and three-quarter views of the ante-mortem face were the most valuable for craniofacial superimposition. However, the poor identification rate achieved using images in frontal view suggests that the MPUB Interpol database would not be optimal for disaster victim identification, and passport-style images do not provide enough distinguishing facial detail. This suggests that multiple ante-mortem images with a variety of facial expression should be utilised for identification purposes. There was no significant difference in success between the manual and computer methods.

Introduction

Facial anthropology involves biological analysis and interpretation of the skull and/or the face and can be applied to identification of the living (biometrics, image analysis) and the dead (facial depiction, craniofacial superimposition, osteology, anatomy) [1]. Craniofacial analysis of the dead can be utilised for single unidentified human remains, multiple victims of disasters and mass graves. In the majority of forensic investigations there will be a suspect in relation to identity; in single forensic cases there may be a missing persons list, in mass disasters there may be a closed list (such as a passenger list for a transport system) or an open list (reported missing by families and/or employers) and for mass graves there may be whole missing populations. In these circumstances there may be ante-mortem data available relating to the suspects and this might include biological profiles (age, sex, stature, ethnic group), personal information (body modifications, identifying marks, clothing, jewellery, hair style, etc.), photographs (ID cards, passport images, family albums or snapshots), hair samples (collected from a hairbrush, etc.), dental records, clinical images and/or medical records [2].

However, much of this ante-mortem data may not be available, for example where the individual did not have a history of dental or medical treatment, or the available data may not be useful, for example where multiple members of the same family are missing (such as in a mass grave or mass disaster) and DNA analysis cannot separate family members. This may be significant in areas/countries with high levels of poverty, low socio-economic status and poor medical/dental practise.

Where ante-mortem images are present, craniofacial superimposition may be effective as it does not require expensive or invasive techniques and is cost and time efficient. Craniofacial superimposition is the process where ante-mortem images are aligned and matched to the skull in order to assess the relationship between the hard and soft tissues of the face. This analysis may allow positive identification, especially where multiple ante-mortem images are available and this has been accepted in international courts as a method of identification [3], [4], [5], [6], [7], [8].

Traditionally craniofacial superimposition has been carried out by forensic anthropologists or anatomists [5], [7], [9] and the techniques incorporate similar anatomical principles [10] and anthropological standards [11] as utilised in facial depiction. Numerous computerised systems have been developed for skull-to-face alignment [8], [12], [13], [14], [15].

The first documented use of craniofacial superimposition for identification in a medico-legal investigation was in 1935 [3], with a particular case involving a Lancastrian GP and the mysterious disappearance of his wife. Dr. Ruxton claimed that his wife had left him for another man, but 2 weeks later two dismembered bodies were found in Glasgow. Police recovered two human heads and over seventy body parts wrapped in newspaper. The newspaper was from a special edition that was distributed only in the area where the Ruxtons lived. Dr. Buck Ruxton had killed his wife and her maid and removed the eyes, noses, lips, skin and teeth to avoid identification of the bodies. The police suspected that the maid and the wife were the victims and a craniofacial superimposition was carried out using ante-mortem images of the two women and photographs of the skulls [16]. The method used in this case was photographic; employing enlargement, measurable objects, anatomical landmarks and craniophore orientation [17]. Known objects in the photographs (e.g. a tiara and a picket fence) were used to enlarge the faces to life size in order to identify Mary Rogers and Isabella Ruxton.

There have been other significant forensic cases where craniofacial superimposition has been utilised for identification of human remains, including the identification of the remains of Josef Mengele [7], [18] and the identification of the victims of the serial killers, Fred and Rosemary West, in the UK [19]. Other significant cases are the Dobkin case [4], the Worlkersdorfer case [5] and the Howick Falls murder case [6]. Detailed reviews of the techniques can be found in the literature [17], [20], [21], [22], [23], [24], [25], [26], [27] and many case studies have also been published [8], [13], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42].

Craniofacial superimposition development has passed through three technological phases: photographic, video and computer-assisted [17], [25]. The photographic technique was pioneered in the 1930s [3], [4], [5], the video technique was developed in the 1970s [43], [44], [45] and the computer-assisted technique was introduced in the 1980s [12], [36], [46].

It is professionally agreed that craniofacial superimposition is of greater value for exclusion than positive identification [17] and that where single facial images are utilised, more than one person may appear consistent with the skull in question. Therefore, forensic practitioners must be well trained in anatomy and anthropology for the effective utilisation of craniofacial superimposition and multiple ante-mortem images of the suspect should be analysed. When evaluating anatomical consistency special attention is paid to the cranial outline, the soft tissue thickness at various anthropometric landmarks on the skull and feature relationships between the skull and the face [47].

It is vitally important to practitioners and law enforcement agencies to establish the level of accuracy of these craniofacial techniques, especially in the case of craniofacial superimposition where this may be utilised in court for positive identification. The credibility of craniofacial superimposition was first established in a study using a database of 52 skulls from the Smithsonian collection [8]. An unknown skull was shown to match an ante-mortem facial photograph, while the four most similar skulls from the collection showed distinct differences to the facial photograph. Another blind study [48] used 3 skulls and compared them to 97 lateral view and 98 frontal view photographs of subjects (including the targets); in total 585 superimpositions. False matches were recorded for 9.6% of the lateral view and 8.5% of the frontal view superimpositions and all 3 targets were correctly matched. The incidence of false matches was reduced to 0.6% when both frontal and lateral view photographs of the same individual were used, suggesting that multiple photographs are optimal to prove or disprove identity by craniofacial superimposition. A further study [17] found that the outline from trichion to gonion in the lateral or oblique view was the optimal portion for personal identification.

Technological advances in the field of facial imaging over the past 20 years have produced new opportunities for research in facial morphology and facial growth assessment [49]. Surface scanning technologies such as laser scanners, photogrammetry cameras and other three-dimensional systems (including three-dimensional digitizers and structured light systems) have advanced craniofacial landmark location [50], [51], [52] and facial anthropometry in both adults and juveniles [53], [54], [55], [56], [57], [58]. These systems have benefited the field of orthodontic and plastic surgery with the ability to record subtle changes in surface morphology, especially important in the monitoring orthodontic treatment, facial growth and maxillofacial intervention [59], [60].

Numerous accuracy studies have been conducted on different laser scanners with regard to the accuracy from three-dimensional capture of faces of the resulting scan measurements when compared with anthropometric measurements from the face [52], [60], [61], [62]. These studies all found that laser scanners are a reliable and fast means of capturing a three-dimensional reproduction of a facial surface whilst maintaining high levels of accuracy with regard to the difference in resulting measurements between the scan and the subject. However, Aung et al. [52] did indicate that accurate location of landmarks and operator skill are important factors in achieving reliable results.

The ability of a forensic practitioner to reliably cite the accuracy of a particular method employed in forensic human identification is highly desirable. The reason for this is not simply for legal reasons but also as a moral and ethical obligation to the victims and their families. This will ensure that if the practitioner is required to provide an expert witness report to the court (with regard to the identification they have assigned to the deceased) the evidence provided is valid with the support of scientific research.

In order to improve craniofacial analysis techniques further rigorous quantitative accuracy studies need to be conducted especially with regard to the scientific method. Improving the accuracy of these techniques and increasing publication of the findings of accuracy studies will allow the use of craniofacial analysis techniques more widely in the field of forensic human identification, especially with regard to disaster victim identification and also increase the acceptance of these methods in a medico-legal investigation.

The aim of this research was to compare the accuracy of manual and computerised craniofacial superimposition techniques and to establish the application of these techniques for disaster victim identification, where there may be a large database of passport-style images, such as the MPUB Interpol database.