Completing the 3-dimensional picture

Orthodontics has a commitment to precision and accuracy similar in concept to physical engineering. Both work in a 3-dimensional (3D) world—the engineer designing and modeling physical products, and the orthodontist diagnosing and treating patients. To achieve the best possible results, both fields have readily adopted and successfully implemented highly advanced mathematical techniques and sophisticated protocols. Today, a notable difference between the 2 fields relates to the data input and analysis/modeling tools that are now available. Technology advances in the past 5 years have started to erode these barriers to safely make human 3D data input as precise and easy to obtain as physical 3D object input. Many are familiar with the advances in 3D computed tomography, specifically cone-beam technology; this article focuses on a sister imaging technology called 3D surface imaging. These 3D systems enable extraoral imaging protocols to be equally precise in providing highly accurate 3D facial surface images for diagnosis, analysis, treatment monitoring, simulation, and outcome evaluation. Orthodontics has a commitment to precision and accuracy similar in concept to physical engineering. Both work in a 3-dimensional (3D) world—the engineer designing and modeling physical products, and the orthodontist diagnosing and treating patients. To achieve the best possible results, both fields have readily adopted and successfully implemented highly advanced mathematical techniques and sophisticated protocols. Today, a notable difference between the 2 fields relates to the data input and analysis/modeling tools that are now available. Technology advances in the past 5 years have started to erode these barriers to safely make human 3D data input as precise and easy to obtain as physical 3D object input. Many are familiar with the advances in 3D computed tomography, specifically cone-beam technology; this article focuses on a sister imaging technology called 3D surface imaging. These 3D systems enable extraoral imaging protocols to be equally precise in providing highly accurate 3D facial surface images for diagnosis, analysis, treatment monitoring, simulation, and outcome evaluation. Orthodontics is distinguished by a commitment to precision and accuracy similar in concept to physical engineering. Both fields work in a 3-dimensional (3D) world—the engineer designing and modeling physical products, and the orthodontist diagnosing and treating patients. To achieve the best possible results, both fields have readily adopted and successfully implemented highly advanced mathematical techniques and sophisticated protocols. Today, a notable difference between the 2 fields relates to the data input and analysis/modeling tools that have historically been readily available to each field. Engineers generally base their work on highly precise 3D data input when 3D surface scanners for capturing inanimate objects became available in the 1980s. Many computer-aided-design/computer-aided-manufacturing (CAD/CAM) software applications were reengineered to make the transition from 2-dimensional (2D) to 3D. As manufacturing firms improved time efficiencies and saved money by streamlining, a paradigm shift became mandatory for the industry. Orthodontists have traditionally based their diagnosis and treatment planning on various 2D radiographs and traditional photographs because conventional 3D input devices for capturing living, breathing human subjects have been prohibitively expensive and complex to use, and often subjected the patient to potentially harmful emissions. Technology advances in the past 5 years have eroded these barriers to safely make 3D human data input as precise and easy to obtain as 3D physical object input (Fig 1). Moreover, to further streamline practice workflow, some practice management and imaging management software applications are being reengineered to efficiently handle and analyze these highly precise 3D data formats. Many are familiar with the advances in 3D computed tomography (CT), specifically cone-beam CT (CBCT) technology; this article focuses on a sister imaging technology called 3D surface imaging. These 3D systems enable the extraoral imaging protocol to be equally precise in providing highly accurate 3D surface images for diagnosis, analysis, treatment monitoring, and outcome evaluation. Although mandibular advancement has a noticeable affect on the appearance of the face, some conditions and orthodontic treatment protocols affect the soft tissues. For example, when a palatal expansion appliance is used, there are often obvious soft-tissue changes to parts of the face (eg, the base of the nose) during treatment. These changes are apparent not only to the orthodontist, but also to the patient. With the readily available soft-tissue capture tools, orthodontists can now accurately document the patient electronically in 3D to evaluate, monitor, and quantify these outcomes (Fig 2). Today, 3D surface data can be registered with 3D CT and CBCT data to provide a thorough view of the patient from the outside in. Although the patient “wow” factor is clear from a marketing perspective for the orthodontic practice, understanding that 3D surface imaging can improve diagnosis and enhance treatment planning is the real push behind the 3D surface imaging (Fig 3). Traditional lateral and frontal photographs have well-known limitations as they attempt to accurately record 3D geometry on a 2D plane.1Honrado C. Lee S. Bloomquist D. Larrabee W. Quantitative assessment of nasal changes after maxillomandibular surgery using 3-dimensional digital imaging system.Arch Facial Plast Surg. 2006; 8: 26-35Crossref PubMed Scopus (68) Google Scholar To work within these recognized limitations, the orthodontic community has established strict photography protocols to extract the best information possible within the constraints of the medium. From its invention, photography has been used to document the human condition. It was readily adopted by the medical community to enhance the patient record. Several variables can cause disparity when comparing 2 apparently similar photographs: distance between camera and subject, camera angle, head position (roll-pitch-yaw orientations), and photography protocol inconsistencies (Fig 4). These problems are further compounded when combining a patient’s lateral photograph with a lateral cephalometric image for analysis. Achieving a repeatable and reliable registration between the 2 images is unlikely, considering all of the variables involved. Sarver2Sarver D.M. Esthetic orthodontics and orthognathic surgery. Mosby, St Louis1998Google Scholar described the limitations of integrating a 2D lateral cephalogram and a 2D lateral photograph for analysis purposes: Usually, facial images and cephalograms are not taken simultaneously, which may result in significant differences in head position and image-magnification discrepancies. Alterations of either the cephalometric profile or photographic image is often required in order to correlate the two images. This “best fit” calibration of the two images provokes questions concerning the validity of quantitative information derived from the imaging systems. Even with the best techniques available for coordinating photographs with corresponding cephalograms, the degree of accuracy may be variable, which is a factor that could affect the validity of the video imaging process. To address these limitations created by head position and magnification, 3D CBCT and 3D surface data sets give the orthodontist digital patient information that is 1:1. In an ideal world, both images would be taken simultaneously to minimize registration errors. Integration of the 2 technologies has several drawbacks because the purpose of the 3D photograph is more than just scraping color texture on top of the CBCT. For example, the patient’s head must be secured in an unnatural orientation to minimize movement during the CBCT scan. The stand-alone 3D surface-imaging system allows the orthodontist to capture multiple surface images during treatment to monitor facial changes and, importantly, the patient’s natural head position to use as the face’s natural orientation for treatment planning (Arnett and Bergman3Arnett G.W. Bergman R.T. Facial keys to orthodontic diagnosis and treatment planning Part I.Am J Orthod Dentofacial Orthop. 1993; 103: 299-312Abstract Full Text PDF PubMed Scopus (353) Google Scholar and Moorrees4Moorrees C.F.A. Natural head position—a revival.Am J Orthod Dentofacial Orthop. 1994; 105: 512-513Abstract Full Text Full Text PDF PubMed Scopus (84) Google Scholar). Because each 3D data set is accurate on its own merit, it is possible to register images that have not been taken at the same time by establishing a facial expression protocol5Idiculla A.J. Harrell W.E. Secchi A.G. Ayala J. Katz S.H. 3-dimensional morphometric facial analysis to determine the effects of altering the occlusal vertical dimension thesis. University of Pennsylvania, Philadelphia2006Google Scholar (Fig 5). Enlow and Hans6Enlow D.H. Hans M.S. Essentials of facial growth. W. B. Saunders, Philadelphia1996Google Scholar reported that “the breadth of the nasal bridge in the region just below the frontonasal sutures does not markedly increase from early childhood to adulthood.” Therefore, this might prove to be a good registration area for growth assessment and treatment effects. Depending on the 3D surface-imaging system used, the process of imaging a patient in 3D can be simpler and take less time than traditional photography. The patient sits in front of the 3D camera, the operator clicks a button, and, from this single 3D data set, any 2D photographic view can be generated—left and right laterals, frontal, left and right obliques, and so on. There is no need to continually reposition the patient in front of the camera; the camera operator simply makes the patient comfortable and relaxed and then acquires each facial expression as required by the practice protocol, whether it is repose, narrow smile, wide smile, or all of them. Because some 3D surface-imaging systems are noninvasive, images can be taken as frequently as needed. Additionally, the imaging speeds from some machine-vision camera-based stereo-photogrammetry systems are so fast that even erratic movement by the youngest child or the most restless patient is not an issue. Once the 3D surface image has been acquired, it becomes the foundation of the visible patient record ready for analysis at any time. Today, several surface-imaging technologies are available to take extraoral 3D images of a patient’s face. Based on commercially available systems, in this article, we will attempt to explain how the raw 3D surface data is generated, how the underlying imaging technologies work from an accuracy standpoint, the importance of acquisition speed, how the technologies are engineered to become a product to handle patient flow, and how to evaluate the company or organization behind the imaging system. To truly take advantage of 3D, the geometry, or shape data, must be an accurate representation of the patient’s anatomy (Fig 6). As in engineering, good data input combined with a solid treatment protocol equals good data output. Typically, depending on the system, there are 2 steps for creating the 3D surface image. First, the geometry is generated, and then the color texture information is applied to the geometry. To mathematically express 3D surface information, the subject’s dimensional face must be converted into a series of coordinates with an x, y, and z definition. These coordinates numerically represent the visible geometry of the patient’s face. Some close-range surface-imaging devices automatically generate a point cloud that is relative to a fixed point, or (0, 0, 0) coordinate. Other close-range surface-imaging devices have adapted the methodology of long-range scanning devices (such as 3D topographic measurement from satellite images) to generate a range map, which represents the relationship between points on the target surface and the location of 1 image sensor. Regardless of method, to adequately cover larger surface areas, such as the face (ear to ear) or the full head, acquisition from several viewpoints is required. To evaluate a system, it is good to understand how the mathematical coordinate system for the entire surface image is generated because it is the foundation for accuracy. In principle, there are generally 2 ways to generate a surface image that is derived from multiple viewpoints. The first is to generate a separate 3D data set (containing its own coordinate system) from each viewpoint, whether a range map or a point cloud, and then stitch them together to produce a new 3D coordinate system. Generating separate data sets for each viewpoint and then stitching them together has historically worked well for data input of inanimate objects since subject motion is not a factor. Unfortunately, this stitching approach does not work well when the subjects are animate, because stitching separate 3D images together to generate a single 3D model of the patient can compromise accuracy.7Singh G.D. Levy-Bercowski D. Santiago P.E. Three-dimensional nasal changes following nasoalveolar molding in patients with unilateral cleft lip and palate: geometric morphometrics.Cleft Palate Craniofac J. 2005; 42: 403-409Crossref PubMed Scopus (85) Google Scholar There are sometimes clues in the 3D data set that this might be the case, such as a visible seam between the 2 original data sets that were stitched together. If there is no seam with this methodology and the capture speed is longer than 1/250 second (equivalent to 4 ms), there might be “smoothing processes” working behind the scene. To eliminate the extra steps—stitching—that increase the probability of creating data errors, the preferable way to generate a 3D surface image derived from multiple viewpoints is to generate a continuous coordinate system by selecting the best quality data for any x, y, and z coordinate from each viewpoint. This is typically accomplished through a 4-ms or less capture speed and sophisticated algorithms for evaluating and analyzing sensor images. Currently, no commercially available 3D surface scanner has been validated to capture accurately on a but some software have been implemented to the 3D data to generate geometry. Because these might not at the it is good to understand how the could affect the accuracy of the data Although there can be a for patients when they their 3D surface when it to geometry for a patient’s there are These the of each for systems and the of a geometry that truly represents the to the of required to represent facial is typically not a From a treatment planning standpoint, to facial diagnosis, analysis, and outcome evaluation. As a most orthodontists and use a 3D surface-imaging have the patients their the face and and behind the The for this is that they to image as of the face as possible the because their diagnosis and treatment to understanding and the soft of the face. Because a patient’s changes from 1 to the there is in the geometry. The of color texture information is the second after the geometry is Depending on the imaging system during this the color information or the texture map is mathematically applied to the shape color geometry to has the most in using 3D input with texture Because the color texture map can a of in the underlying geometry, the first by most digital when a 3D data is to the color texture so that they can the foundation geometry. Some 3D photography systems are used as a imaging from 1 location to to of data from patient the geometry of the patient’s face is the for these for of and This compromise can be because color and texture or visible on the texture have no on the accuracy of the geometry. It is imaging the surface of a subject, the speed of capture is In to when of geometry are combined with the surface geometry, the 3D surface image has a single of x, y, and z coordinates to the geometry and the color Because patients are living, and to the speed of acquisition must be fast to motion is to affect traditional camera provide speeds for a For a the speed is the to use, because less time for subject movement to the speeds will all working speed for with a is to camera and all but need or with Scholar to Scholar difference in real time between 1/250 of a second and of a second may not but when in photographic It the difference between the motion and a In a based 3D surface-imaging system, are to image the It is not for each camera to on its own within the 1/250 time (equivalent to 4 all in the 3D system must within the 1/250 time to capturing the subject at the same time. With this could to to or to of a or With some systems, for or the model software is to for the time by in and in the data, as well as the geometry when the raw geometry data is or These can when the accuracy of the geometry. Even each of these techniques and can from to are In imaging the scanner was of the first 3D surface-imaging technologies available. Depending on the it geometric to determine the surface coordinates of the target As the the scanning a or the surface area of the target The object the which is then at a distance from the using the x, y, and z coordinates of a surface point are Although the 3D model generated might be accurate on a a single human face of from top to and each is taken To achieve any accuracy in capturing the face, can take to this of time adequately for applications such as engineering, quality and technologies are to use on living, breathing S. S. surface for analysis of the cleft Palate Craniofac J. 2006; Google Scholar imaging technologies and The of such as or the subject, simultaneously the subject with a camera that is with the of the the of the the photograph is by the software to generate the shape texture information is registered with the point range Although this is for capture areas, the is to generate an accurate 3D model of a human subject’s face from to because it a to the of the these images must be taken in to a from 1 viewpoint a from image capture the acquisition for human can be to data accuracy. The is a sophisticated software approach based on the of 2 of the same at a distance similar to the distance between a to a and to record sophisticated algorithms to and surface between the 2 photographs and generate a 3D If the system a point cloud, the underlying software must the position of each camera sensor relative to the this is a target during the calibration The on the surface the algorithms with the base information required to accurate geometry. Once the 3D geometry model has been the software the color texture information the There are two for and The approach its on the natural or on the surface of the target To image the face, the approach on its to on the and so on. Because this approach highly on the of the it typically to that there is surface to the of using 3D image J In Scholar As a the system operator must to the conditions to minimize and address the that result in to and several in a to image a patient. In an approach its on the natural or on the surface of the target object and this natural with a to give the algorithms as information as possible to the geometry (Fig Typically, is more to in since the system full of subject during image this also of relates to color algorithms can capture and since the in and provide surface information to the x, y, and z with the a system can image the of has 2 accuracy with to the physical of an and repeatable accuracy between a 3D imaging system, it is a good to for an accuracy by an it is to use an inanimate object as the subject when stitching data together because there is no make and of from J 2005; Scholar For the 3D surface image to have any the geometry generated must the of the surface to a of a of selecting a 3D surface-imaging focuses on the use of the and how it will work in a With such as and it would be to evaluate the and In most is so the the area required to the system, the From a patient perspective and a standpoint, there are a to evaluate when the system in the First, it is to understand the entire not just the but the required for the the and any because some systems a behind the patient. these systems typically on them in such as a the of the system and of Many have systems that of patient protocol is the 3D surface image is to be accurately registered to a or traditional CT image. It is to understand the entire process from the system and capturing the data to the 3D image and the in a practice management To understand will be with the of the it is the system in use in a practice and to its 3D systems calibration of the sensor use, so that the can achieve on each Some to are calibration it and can it be by the data capture the patient for a of time and how images are taken to obtain the 3D data process automatically or how does it take the 3D image is ready for and with a practice management system. If an orthodontist to image all or most patients during not just a for an the 3D surface-imaging system must be reliable and to To a for the of a system, it is good to understand its and its evaluating systems, the of in the system could have an on time. For example, medical machine-vision into its system (Fig These are generally of a quality and than in and are for and applications for several years and have a to be to the In camera capture speeds are by the speed with the If the in the 3D system are with the camera speeds capture the image any than the speed with to The of Scholar speed is to fast 1/250 is not we and is of the need to at or to patients an orthodontist, it does not to simply has and typically use a orthodontists by from their on their and the of the and during the treatment evaluating 3D surface-imaging system an orthodontist take a similar a at the base such as the of 3D systems to and their location in to the of these the working with the system in a practice or whether the organization is to to software from software and the of are orthodontists and the record to This is because a base and technology If the base of or how will this affect the orthodontist in product that the organization can its commitment to the orthodontic can cover all imaging system points in this article and the commitment to with practice management systems and 3D software applications to specifically orthodontists with treatment monitoring, and outcome evaluation. With 3D input devices and software now for human data, orthodontists can problems using 3D just their With precise shape data of each orthodontists and can these at any time to analyze or the patient and continually monitor the during treatment. With the to continually monitor and quantify can be more in treatment with the range of new and treatment techniques a less outcome on their patients. 3D surface-imaging system 3D CBCT by surface information to the Because it is noninvasive, patients can be at during treatment.