Creation of the Virtual Patient for the Study of Facial Morphology




The author provides an overview of the new imaging technologies that allow the practitioner to accurately capture the patient’s soft tissue facial morphology and underlying bones and teeth, including details of dental model integration. This article describes how a virtual patient is created and manipulated and the practical use of this technology. It takes the quantification of the 3D surface further by proposing a reference framework of landmarks of craniofacial structure that can be used for comparison of surgical change, growth, gender, and phenotype.


For clinicians, the head and neck region is a complex area to work in. Multiple craniofacial anomalies and dentofacial deformities are treated on a daily basis. With better techniques and less-invasive procedures, the face is also an area where cosmetic procedures are performed routinely.


Unlike other areas of the body, the face is unique and gives an individual identity; it relays who we are, where we come from, and what we think or feel. The heavy emphasis on facial beauty and aesthetics has recently shifted the way orthodontists, cosmetic dentists, oral and maxillofacial surgeons, and plastic and reconstructive surgeons look at the human face. For any given face, diagnosis and treatment start from the outside in, meaning from the soft tissue surface to the hard tissue structures. Some in the profession have termed this procedure the soft tissue paradigm.


Many craniofacial teams still rely on the traditional methods for diagnosis and treatment planning. However, this should not be the case. With modern techniques and technologies, a realistic 3-dimensional capture of the face to which texture and color are added is achievable and should be of prime importance to all craniofacial researchers and clinicians.


The advances in technology allow us to reconstruct the human face layer by layer, integrating bone, teeth, and skin together to generate a volumetric model of the actual patient. This article describes how the virtual patient is created and manipulated to reach the best treatment outcome. In addition, it also proposes a method of analysis of the human form.


3-dimensional diagnostic records


In the image capture of the facial form, 3 primary areas are easily reproduced with current techniques. These are



  • 1.

    The surface texture of the face


  • 2.

    The skeletal structure of the head and neck


  • 3.

    The teeth and its corresponding occlusion.





Surface acquisition technology


Traditionally, extraoral photographs are taken to represent the face before treatment. This photograph serves as a baseline record and also as a planning material. Careful orientation of the head and control of the lens magnification are important for the clinician; however, these photographs are only 2 dimensional, and, even though many views are usually taken, they do not reflect a 3-dimensional reality of the result.


There are a multitude of 3-dimensional surface acquisition systems that are described by Kau and colleagues in other articles in the literature. The most popular technique is stereophotogrammetry. This system is a technology that allows the capture of volumetric surfaces. Usually, 2 or more cameras on each side work as stereo pairs, and a method of triangulation using complex algorithms are used to merge the obtained pictures and create a volume.


The 3dMD system (3dMDface, Atlanta, GA, USA) was used in this article. This machine is equipped with 6 cameras, 3 on each side, that acquire volume, texture, and color. Its capture time is 1.5 milliseconds. The system produces 1 continuous point cloud from the 2 camera viewpoints and generates a 3-dimensional composite model. Texture and colors are layered over the model. The cameras can only read the well-illuminated anatomic parts of the face; therefore, the submental area, the inside of the nostrils and ears, and the hairline sometimes show as voids. Well-groomed facial hair, on the other hand, does not seem to affect the scanning procedure. Images are stored as proprietary software versions as ∗.tsb files from the 3dMD files. The 3dMD system presents multiple advantages. Other than the speed of data acquisition and the reported geometric accuracy (<0.2 mm root mean square or better) (depending on exact configuration), it allows the capture of the face in natural head position (NHP).


NHP is considered to be the most natural physiologic and anatomic orientation of the head. This position is internally regulated by each individual and results from a combination of visual, sensory, and postural reflexes. This position is usually achieved by asking patients to look at a distant point in the horizon or at their own reflection in a mirror. NHP has been proven to be clinically reproducible and has been used in all studies of the author.


Another important component of surface texture capture is the facial pose. It is important that a technique be used to capture a reproducible pose. A previously published study has shown that facial pose is reproducible up to 0.85 mm over a 3-day period. This possibility is remarkable considering the variation of facial form that can exist between 2 periods.


The techniques used require the patient to be seated on an adjustable stool preventing strain to the muscles of the neck. The patients are asked to level their eyes to the horizontal line and to align the midline of their faces with the vertical line drawn on a mirror facing them. The subjects are told to swallow hard and to keep their jaws in a relaxed position just before the images are taken. The 3-dimensional image is captured almost instantaneously with a click of a button.




Surface acquisition technology


Traditionally, extraoral photographs are taken to represent the face before treatment. This photograph serves as a baseline record and also as a planning material. Careful orientation of the head and control of the lens magnification are important for the clinician; however, these photographs are only 2 dimensional, and, even though many views are usually taken, they do not reflect a 3-dimensional reality of the result.


There are a multitude of 3-dimensional surface acquisition systems that are described by Kau and colleagues in other articles in the literature. The most popular technique is stereophotogrammetry. This system is a technology that allows the capture of volumetric surfaces. Usually, 2 or more cameras on each side work as stereo pairs, and a method of triangulation using complex algorithms are used to merge the obtained pictures and create a volume.


The 3dMD system (3dMDface, Atlanta, GA, USA) was used in this article. This machine is equipped with 6 cameras, 3 on each side, that acquire volume, texture, and color. Its capture time is 1.5 milliseconds. The system produces 1 continuous point cloud from the 2 camera viewpoints and generates a 3-dimensional composite model. Texture and colors are layered over the model. The cameras can only read the well-illuminated anatomic parts of the face; therefore, the submental area, the inside of the nostrils and ears, and the hairline sometimes show as voids. Well-groomed facial hair, on the other hand, does not seem to affect the scanning procedure. Images are stored as proprietary software versions as ∗.tsb files from the 3dMD files. The 3dMD system presents multiple advantages. Other than the speed of data acquisition and the reported geometric accuracy (<0.2 mm root mean square or better) (depending on exact configuration), it allows the capture of the face in natural head position (NHP).


NHP is considered to be the most natural physiologic and anatomic orientation of the head. This position is internally regulated by each individual and results from a combination of visual, sensory, and postural reflexes. This position is usually achieved by asking patients to look at a distant point in the horizon or at their own reflection in a mirror. NHP has been proven to be clinically reproducible and has been used in all studies of the author.


Another important component of surface texture capture is the facial pose. It is important that a technique be used to capture a reproducible pose. A previously published study has shown that facial pose is reproducible up to 0.85 mm over a 3-day period. This possibility is remarkable considering the variation of facial form that can exist between 2 periods.


The techniques used require the patient to be seated on an adjustable stool preventing strain to the muscles of the neck. The patients are asked to level their eyes to the horizontal line and to align the midline of their faces with the vertical line drawn on a mirror facing them. The subjects are told to swallow hard and to keep their jaws in a relaxed position just before the images are taken. The 3-dimensional image is captured almost instantaneously with a click of a button.




Cone beam computerized tomography


To create the virtual patient, the 3dMD image is combined with an imaging source that captures the skeletal tissues. There are 2 main methods for skeletal hard tissue capture, one being the traditional spiral computerized tomography and the other the cone beam computerized tomography (CBCT). Because of irradiation concerns for routine patient care, a lower-dosimetry emitting machine is recommended. Like the surface acquisition system, the subjects are placed in NHP when the CBCT images are taken . CBCT allows the acquisition of 3-dimensional images of the skull and teeth. It can also capture soft tissue surfaces, but the rendered quality of soft tissue is not ideal. Furthermore, the extent of the field of view does not include the cranium or go past the cephalometric point glabella. Hence, CBCT images need to be combined with stereophotogrammetry.


The cone beam produces a more-focused beam and much less radiation scatter compared with the conventional fan-shaped computed tomographic (CT) devices. The estimated radiation dose is between about 60 and 1000 μSv. Capture time is less than 20 seconds during which the machine revolves once or twice around the patients’ head. The machine is equipped with a conventional low-radiation x-ray tube that focuses the beam on a flat-panel detector or charge-coupled device. It is estimated that the total radiation involved in a CBCT is equivalent to 20% of that of conventional CT and corresponds to the dose generated by full-mouth periapical radiographs. Raw data are stored in Digital Imaging and Communications in Medicine (DICOM) file formats. Further reading on the device may be found in the literature.




Dental model integration


Traditionally, dental stone casts have been used by clinicians to visualize the occlusion, perform analysis, and fabricate appliances. The introduction of 3-dimensional digital casts has tremendously decreased chairside time and reduced storage space in orthodontic offices. Digital dental casts have been reported to provide an accurate and reliable representation of the dentition. These virtual models can be produced in 2 ways. In the first method, dental impressions are scanned and the generated models can be viewed with the adequate software. Some investigators have tried to integrate these dental images in the CBCTs, stating that raw dental arches derived from CBCT itself can be inaccurate because of metallic scatters.


The new challenge is then to integrate these 3-dimensional dental models in the virtual patient reconstruction. The technique used by Gateno and colleagues allows an accurate incorporation of the teeth in the skull but causes soft tissue distortion because of the jig they use. Swennen and colleagues report an interesting approach to merge the models with the hard tissue but use 3-stage scans to achieve their goal, which increased the radiation dose in the patient. The second method is to use appropriate software to create study models directly from the CBCT. A good example is InVivo Anatomage (Anatomage, San Jose, CA, USA) that can create study models from CBCT scans taken by a Galileos cone beam scanner with a field of view of 15 × 15 × 15 cm and a voxel resolution of 0.125 mm. The CBCT images are then electronically sent via a secure Web site to the company Anatomage in a DICOM format. This approach significantly cuts down the record acquisition steps because no impressions are required and the teeth are already integrated in the skull.


Some preliminary works on the direct rendering of dental casts directly from CBCT DICOMs have shown some promise. However, these representations are not enough for appliance fabrication. A study using the Little’s index compared CBCT study models with OrthoCAD models (CADENT, Fairview, NJ, USA) and found them to be as accurate as the digital models for measuring overjet, overbite, and crowding. Moreover, CBCT models offer diagnostic information on bone level, root orientations, and temporomandibular situation. The advantages of digital models are numerous. Recovery of the right dental occlusal relations allows a more-accurate treatment diagnosis, planning, and adequate appliance design. Canine impactions or multidisciplinary treatments, especially those involving surgery, can benefit from this technology.




Creating the virtual patient


3dMDvultus Software


Some software platforms permit the fusion of the surface texture (acquired by the 3dMD system) with the hard tissue (CT/CBCT/digitized dental study models). One such platform is the 3dMDvultus system. Because both soft and hard tissue images are taken in NHP, fusing the CBCT-generated soft tissue surface with the stereophotogrammetric image allows total visualization and manipulation of the created virtual head.


The steps needed to merge soft tissue to hard tissue in this software are described by the manufacturer and consist of the following:



  • 1.

    The patient-acquired surface images and CBCT results are loaded in the software, and segmentations into surfaces are created.


  • 2.

    After unlocking rotation and translation functions, the operator manually fits the soft tissue generated by the CBCT on the 3dMD surface.


  • 3.

    The registration is surface based. The operator selects anatomically stable surfaces, and the software refines the initial manual registration.


  • 4.

    Once the surfaces are registered, it becomes possible to visualize volumes or cuts in different degrees of transparency.



The following section illustrates the sequence in which data are acquired and the virtual model created for a patient who presented at the Craniofacial Clinic at the Department of Orthodontics, University of Alabama at Birmingham.


The sequence used is as follows:


Feb 8, 2017 | Posted by in General Surgery | Comments Off on Creation of the Virtual Patient for the Study of Facial Morphology

Full access? Get Clinical Tree

Get Clinical Tree app for offline access