Sven De Greef

Computerized craniofacial reconstruction: Conceptual framework and review

When confronted with a corpse that is unrecognizable due to its state of decomposition, soft-tissue mutilation or incineration, and if no other identification evidence is available, craniofacial reconstruction (CFR) can be a useful tool in the identification of the body. Traditional methods are based on manual reconstruction by physically modelling a face on a skull replica with clay or plasticine. The progress in computer science and the improvement of medical imaging technologies during recent years has had a significant impact on this domain. New, fast, flexible and computer-based objective reconstruction programs are under development. Employing the newer technologies and permanently evaluating the obtained results will hopefully lead to more accurate reconstructions, beneficial to the added value of CFR methods during crime-scene investigations. A general model-based workflow is observed, when analysing computerized CFR techniques today. The main purpose of this paper is to give an overview of existing computer-based CFR methods up to date defined within a common framework using a general taxonomy. The paper will also discuss the various alternatives and problems which arise during the process of designing a CFR program.

Semi-automated Ultrasound Facial Soft Tissue Depth Registration: Method and Validation

A mobile and fast, semi-automatic ultrasound (US) system was developed for facial soft tissue depth registration. The system consists of an A-Scan ultrasound device connected to a portable PC with interfacing and controlling software. For 52 cephalometric landmarks, the system was tested for repeatability and accuracy by evaluating intra-observer agreement and comparing ultrasound and CT-scan results on 12 subjects planned for craniofacial surgery, respectively. A paired t-test evaluating repeatability of the ultrasound measurements showed 5.7% (n = 3) of the landmarks being significantly different (p < 0.01). US and CT-scan results showed significant differences (p < 0.01) using a Wilcoxon signed rank test analysis for 11.5% (n = 6) of the landmarks. This is attributed to a difference in the volunteers head position between lying (CT) and sitting (US). Based on these tests, we conclude that the proposed registration system and measurement protocol allows relatively fast (52 landmarks/20 min), non-invasive, repeatable and accurate acquisition of facial soft tissue depth measurements.

Bayesian estimation of optimal craniofacial reconstructions.

Forensic craniofacial reconstruction (CFR) aims at estimating the facial outlook associated to an unknown skull for victim identification. Computerized CFR techniques are essentially a virtual mimicking of manual CFR techniques and all share the same conceptual model-based framework. We propose a fully automated Bayesian based statistical framework estimating the most probable face, according to a known craniofacial model (CFM), given the, possibly inaccurate, skull data. A multivariate Gaussian distribution is assumed for the shape parameters of the CFM, only allowing face-like solutions. The CFM is improved by encoding tissue depth differently as an extra value for 52 landmarks on the face and by incorporating gray-valued texture information. A fully automated and consistent technique is obtained by the use of an implicit target skull representation (TSR). The most plausible face associated to the target skull is calculated using an expectation-maximization procedure that is robust to small (noise) and/or gross errors (outliers). A clinical database of 12 individuals is used for simulating realistic reconstruction scenarios. Validation is performed in terms of reconstruction accuracy and recognition success. Within the same EM reconstruction framework, the proposed procedure is compared to alternative reconstructions using different target skull representations and different CFMs incorporating various amounts of covariance. The results indicate that the proposed CFM performs better than the other models. Furthermore, the use of the implicit TSR generates more consistent and better results compared to a realistic landmark based skull representation. Finally, these results also confirm that the Bayesian framework formulation is indeed robust against noise and outliers in the skull data.

A spatially-dense regression study of facial form and tissue depth: Towards an interactive tool for craniofacial reconstruction

Forensic Craniofacial Reconstruction (CFR) is an investigative technique used to illicit recognition of a deceased person by reconstructing the most likely face starting from the skull. A key component in most CFR methods are estimates of facial soft tissue depths (TD) at particular points (landmarks) on the skull based on averages from databases of TD recordings. These databases vary in their method of extraction, number and position of landmarks (usually sparse <100), condition of the body, population studied, and sub-categorization of the data. In this work a new dataset is presented in a novel manner based on 156 CT scans using a spatially-dense set (∼7500) of TD recordings to allow for a complete understanding of TD variation interpolating between typical landmarks. Furthermore, to unravel the interplay between soft-tissue layers, skull and facial morphology, TD and Facial Form (FF) are investigated both separately and combined. Using a partial least squares regression (PLSR) analysis, which allows for working with multivariate and spatially-dense data, on metadata of Sex, Age and BMI, different significant patterns on TD and FF variation were found. A similar, but with TD and FF combined, PLSR generated a model useful to report on both, in function of Sex, Age and BMI. In contrast to other datasets and due to the continuous nature of the regression there is no need for data sub-categorization. In further contrast, previous datasets have been presented in tabulated form, which is impractical for spatially-dense data. Instead an interactive tool was built to visualize the regression model in an accessible way for CFR practitioners as well as anatomists. The tool is free to the community and forms a base for data contributions to augment the model and its future use in practice.