Study on the criteria for assessing skull-face correspondence in craniofacial superimposition
Introduction
Craniofacial superimposition (CFS) [1] is one of the approaches used in craniofacial identification [2], [3]. It involves the superimposition of a skull (or a skull model) over a number of ante mortem images of an individual and the analysis of their morphological correspondence. Since the first documented use of CFS for identification purposes [4] the technique has been undergoing continuous improvement. Although the foundations of the CFS method were laid by the end of the nineteenth century [5], [6], the associated procedures evolved as new technologies became available. As a result, distinct approaches have developed: photographic, video, and computer-aided superimposition [1], [7], [8]. Regardless of the applied technology, some authors have recently described three different stages for the whole CFS process [8], [9]: i) the acquisition and processing of the skull (or skull 3D model) and the ante mortem facial images together with the anatomical landmarks; ii) the skull-face overlay (SFO), which focuses on achieving the best possible superimposition of the skull and a single ante mortem image of the missing person. This process is repeated for each available photograph, obtaining additional overlays. Skull-face overlay thus corresponds to what traditionally has been known as the adjustment of the skull size and its orientation with respect to the facial photograph [10], [11]; iii) decision making. The degree of support for being the same person or not (exclusion) is determined by studying the relationship between the skull and the face based on the superimpositions achieved in the latter SFO stage: the morphological correlation, the match between the corresponding landmarks according to the soft tissue depth, and the consistency between asymmetries.
Although its reliability is still open to discussion, CFS has been employed by both the forensic anthropology community and law enforcement in the identification of unknown persons. It is used together with other techniques or on its own when there is insufficient information available to apply other techniques. CFS has been used for almost a century [2], contributing to the process of many identification cases, especially in scenarios like mass disasters [12], terrorism [13], missing person’s identification [14], common grave investigation [15], and historical cases [16]. There is lack of protocols and standards in the application of the technique and varying information concerning its reliability [17], [18], [19]. The ‘New Methodologies and Protocols of Forensic Identification by Craniofacial Superimposition’ (MEPROCS) project [20] aims to develop “a common framework to allow the extensive application of the CFS technique in practical forensic identification scenarios commonly tackled by European scientific law enforcement, providing an objective evaluation of the forensic identification results achieved by CFS, avoiding particular assumptions that could bias the process”.
To this aim the MEPROCS international consortium, composed of 26 institutions including research centres, universities, police forces and international associations, set out to produce a set of work packages, meetings and inter-lab experimental studies. The latter are intended to provide quantitative and objective data that could support discussions and facilitate decision making processes in an unbiased way. In the first study [21], [22] 26 participants from 17 different institutions in 13 countries were asked to deal with 14 CFS identification scenarios, some of them involving the comparison of multiple candidates with multiple unknown skulls. In total, 60 SFO problems, divided into female and male sets, were analysed. Participants followed their own methodologies and employed their own particular technologies. The data obtained from this large study was a key result leading to an international agreement on the first standard in the field. It includes good and bad practices, sources of error and uncertainties, technological requirements and desirable features, and finally a common scale for the craniofacial matching evaluation [22].
However, that study and the subsequent conclusions mainly focused on the process of superimposing the skull over the facial photograph, the aforementioned SFO stage [10]. Although it also deals with the relation between the quantity and quality of the materials (skull, ante mortem photographs) and the degree of support for a given identification decision, it did not cover the analysis of the skull and face anatomical relationship. One of the main reasons for this limitation is due to the different SFOs achieved by the participants in each single case. A visual inspection of participants’ results clearly shows a important variability in the superimpositions achieved which, biased the following skull-face relationship assessment stage. In addition, as participants were asked to follow their own methodology, the set of anatomical criteria was different for each participant.
As a consequence, the MEPROCS consortium designed the current study which aims to analyze the subjectivity and discriminative power of the different criteria for assessing the skull-face correspondence either proposed in the literature or by any of the MEPROCS partners. The following four tables (Table 1, Table 2, Table 3, Table 4) group all the craniofacial assessment criteria MEPROCS partners considered relevant. The 65 criteria represent an exhaustive list of the criteria employed and described in the most important studies in the field [1], [17], [18], [23], [24], [25]. These criteria, to be used in the assessment of the consistency between the skull and the face, are organized in four different groups analysing anatomical criteria such as lines, landmarks and the corresponding soft tissue thickness, the concordance between the outlines of the face and the cranium, and positional relationship of specific facial and cranial features.
This novel study is expected to provide important insights to better understand: i) which are the most and least discriminative criteria; ii) which criteria depended more on the expert and which criteria are more independent, i.e. less subjective. Those criteria that are determined to be more discriminatory could be included as a recommended standard for CFS.
Section snippets
Material and methods
The dataset used in this study consisted of 18 different CFS problems, some datasets included more than one image of the same subject (24 SFO in total). Three-dimensional skull models were obtained from patients whose head was scanned with a Cone Beam Computed Tomography (CBCT) in orthostatic position. The patients gave their consent to use their clinical data and their scans were anonymized for this study.
In addition, frontal and lateral photographs were taken of the same patients to create a
Results
With the aim of providing a feasible forum for discussion this study focused on the analysis on one scenario, where all the data (participants and CFS cases) are considered simultaneously (see Section 2.2, 1.a). In addition, in segments of this manuscript, the second scenario where the data were divided in two different sets according to the view of photograph (frontal and lateral) is addressed. The remainder scenarios provided similar results (scenarios 1.b and 1.c) or resulted of minor
Discussion and conclusions
The purpose of this experimental study was to provide quantitative and objective data for discussion in order to attempt to reach an international agreement (among MEPROCS partners) of a set of recommended craniofacial assessment criteria.
Although we have examined all the material produced according to the three different statistical analyses and the different types of correlation, the results and discussion included in this manuscript focus only on the most relevant results. In particular, we
Acknowledgements
We would like to thank all the participants that give us the permission to work with both their head scan and facial photographs and Drs. Luca Contardo and Domenico Dalessandri to the support provided during images acquisition and head scanning. We are also grateful towards the Universitary Hospital of Trieste and Ortoscan for supporting this research.
References (33)
- et al.
Superimposition and reconstruction in forensic facial identification: a survey
Forensic Sci. Int.
(1995) - et al.
Past, present, and future of craniofacial superimposition: literature and international surveys
Leg. Med.
(2015) - et al.
Computer vision and soft computing for automatic skull-face overlay in craniofacial superimposition
Forensic Sci. Int.
(2014) - et al.
Study on the performance of different craniofacial superimposition approaches (I)
Forensic Sci. Int.
(2015) - et al.
Study on the performance of different craniofacial superimposition approaches (II): best practices proposal
Forensic Sci. Int.
(2015) - et al.
Evaluation of anatomical consistency in craniofacial superimposition images
Forensic Sci. Int.
(1995) A study on national differences in identification standards for Chinese skull-image superimposition
Forensic Sci. Int.
(1995)- et al.
Cranio-facial morphanalysis: a new method for enhancing reliability while identifying skulls by photo superimposition
Forensic Sci. Int.
(2001) - et al.
Comparison between effective radiation dose of CBCT and MSCT scanners for dentomaxillofacial applications
Eur. J. Radiol.
(2009) Craniofacial superimposition
Craniofacial identification: techniques of facial approximation and craniofacial superimposition
Medico-Legal Aspects of the Ruxton Case
The Bertillon System of Identification
Computer-assisted photographic superimposition
J. Forensic Sci.
Forensic identification by computer-aided CFS: a survey
ACM Comput. Surv.
Cited by (14)
Facial soft tissue thickness differences among different vertical facial patterns
2020, Forensic Science InternationalAutomatic skull-face overlay and mandible articulation in data science by AIRS-Genetic algorithm
2020, International Journal of Intelligent NetworksCraniofacial photographic superimposition: New developments
2019, Forensic Science International: SynergyCitation Excerpt :Statistical analysis and interpretation of results revealed patterns of criterion usage, expert proficiency, criterion subjectivity and criterion discriminative power. Using the MEPROCS recommended best practices and assessment criteria, gleaned from previous studies, 12 participants of varying experience addressed additional superimposition scenarios (Ibáñez et al., 2016) [27]. Statistical analysis of results indicated that following the MEPROCS recommendations improved performance.
Hierarchical information fusion for decision making in craniofacial superimposition
2018, Information FusionCitation Excerpt :These distances can be checked using the skull-face overlay in existing studies relating to soft tissue thickness in different human populations [19]. MEPROCS work group also discussed and quantitatively analyzed these criteria for the evaluation of the morphological skull-face correspondence, providing a set of the most discriminative and easy to assess criteria [24]. Our long term goal is to automate the whole decision making process by modeling the most relevant criteria within the previous four families using CV and SC techniques.
Genetic algorithms for skull-face overlay including mandible articulation
2017, Information SciencesCitation Excerpt :This dataset will allow us to analyze the performance of our methods, following an objective and reliable procedure. Previous works in collaboration with forensic experts remark the importance of objective experimental analysis to validate the results and improve the confidence of the CFS technique [28]. In this proposal we aim to extend the ground truth dataset of [26] following a similar procedure.
MEPROCS framework for Craniofacial Superimposition: Validation study
2016, Legal MedicineCitation Excerpt :We find two possible explanations of the lack of correlation for these criteria. The two previous proposals for achieving an optimal and unbiased SFO [19] and for the analysis of the criteria for assessing skull-face correspondence in CFS [14] were really important contributions in the field. However, the conclusions drawn in the previous studies were influenced by the materials employed (i.e. cone-beam CTs that lacked the upper part of the skull and presented additional inconveniences), as pointed out by the authors., Therefore, more thorough studies should be developed.