Quantitative Assessment Protocol for Facial Soft Tissue Volumetric Changes with Stereophotogrammetry

The assessment of facial soft tissue changes is important in dentistry because facial soft tissues are an essential part of esthetics and function and are considered an integral part of treatment planning across dental disciplines. Soft tissue evaluation allows clinicians to predict the impact of therapeutic interventions-such as orthodontics, prosthodontics, and orthognathic surgery-on facial appearance and balance¹. Relying solely on skeletal or dental measurements can result in compromised treatment outcomes, as the achievement of ideal occlusion and/or skeletal relationships does not always translate to balanced or desired facial soft tissue changes^2,3.

Soft tissue analysis is essential for individualized treatment planning as it helps identify baseline facial characteristics, anticipate the effects of growth and aging, and predict how dental, orthodontic, or surgical interventions will alter facial contours and structural support^4,5. Specifically, changes in lip support, facial convexity, and lower facial height are directly influenced by dental and skeletal modifications, and these changes can significantly affect patient satisfaction and the perceived aesthetic outcome^2,4. Accurate assessment of soft tissue changes is crucial for achieving optimal results in esthetic zones, particularly in implant and restorative dentistry, where the preservation or augmentation of soft tissue volume is necessary to maintain a natural appearance and function⁶. Modern three-dimensional imaging and scanning technologies have improved the precision and reproducibility of soft tissue measurements, facilitating better diagnosis, treatment planning, and outcome evaluation^7,8,9.

The gold standard for soft tissue assessment is serial 3D imaging to quantify soft tissue volume changes, supplemented by clinical examination and patient-reported symptoms¹⁰. The most accurate and reproducible approach is serial 3D stereophotogrammetry or laser surface scanning, which allows for precise measurement of soft tissue volume changes over time^10,11,12. Previous publications have described quantitative methods for the assessment of facial soft tissue changes in the oral and maxillofacial literature, though reported outcomes remain highly variable^{10,11,12,13,14,15,16,17}. However, in most of the previous studies, details regarding the steps involved are not included, and the reproducibility of the results is often not reported. Moreover, there is no video demonstration of the manual part of the process, which would be practically very valuable to clinicians and researchers who are interested in conducting soft tissue assessment. This video demonstration showcases in detail a methodology for conducting a semi-automated, high-accuracy, and highly reproducible 3D volumetric change assessment using facial surface imaging.

This protocol was conducted in accordance with the ethical guidelines established by the Institutional Review Board of the National Institutes of Health (NIDCR IRB #16-D-0040). Informed consent was obtained from all subjects under an NIH IRB-approved protocol (NCT02639312), permitting the use of the3D facial photographs in research-related publications. All imaging was performed at the NIH Dental Clinic with the use of a 3D imaging system¹⁸. See the Table of Materials for details regarding the software used in this protocol. Explicit consent was obtained for the use of the 3D facial photographs used in this protocol.

1. Image preparation

1. The system captures three images: Right, Front, Left, and automatically processes these into a single 3D image. If auto-stitching fails, ensure the images were taken in the specified order and that positioning was correct (camera angle, open eyes, closed mouth). If the images appear to be properly captured, place manual landmarks as instructed by the software, and once satisfied, click retry stitch.
2. To manipulate the image without changing the spatial orientation, use the tools indicated with an eye icon in the left menu: move, rotate, tilt, zoom.
3. Remove unnecessary areas from the image using Box Crop to quickly eliminate everything outside the selected rectangular frame.
   NOTE: For more precise editing, use the Paint Area Selection, which allows for the manual selection of areas for removal, with the option to adjust brush size for detailed refinement.
4. Save the edited image using the save option from the File menu.

2. Baseline image registration to the 3D axis grid

NOTE: The baseline image is defined by the study design. This image is a reference from which the individual subject's subsequent images will be registered. Midline symmetry is established as follows:

1. Select "Axis Grids" from the top toolbar options to display three-dimensional axis grids for all planes.
2. Click on Snap View from the top toolbar to square the image to the nearest 90° frontal view.
3. From the left-side toolbar, select Spin Active Surfaces (cube icon) and rotate the image until it is evenly bisected by the vertical (Y) axis.
4. Using the Paint Area Selection tool from the "Tools" menu of the "Area" option in the top toolbar, drag the brush across the frontal surface of the image by holding down the left mouse button and moving the cursor to highlight the face
5. Once selected, click Find Symmetry from the "Surface" menu in the top toolbar to automatically align the image along the vertical axis through its center
6. Use Clear Area from the "Select" option from the "Area" menu in the top toolbar to deselect the region after.
7. To correct image rotation and establish front-to-back orientation for registration, begin by displaying the image in two viewports by selecting the Show One Surface in Each of Two Viewports option from the top toolbar.
8. Use the Spin tool (eye icon) from the left-side toolbar and Snap function from the top toolbar to obtain a lateral view. In the lateral viewport, use Roll Active Surfaces (cube icon) from the left-side toolbar to adjust the image so that the head position is aligned vertically with the grid. Next, select Pan Active Surfaces (cube icon) from the left-side toolbar and move the image within the lateral view until it is centered on the main vertical axis. Be sure to save the registered image.

3. Landmark annotations

NOTE: A soft tissue landmark-based registration was selected over a surface-based method to better account for variation in surface availability, which is limited by image quality and restricted to regions unaffected by surgery. The sequence of landmark annotations should be the same for each image. If possible, annotate a patient-specific landmark, such as a scar or mole, that lies outside of the region affected by surgery. The landmark selection should depend on the area of interest for each assessment.

1. To follow this protocol, use the following landmarks (see Figure 1):
   1. Medial canthus/Endocanthion (en): the inner commissure of each eye fissure. Annotate the left first, then right. Verify correct placement in the frontal view.
   2. Lateral canthus/Exocanthion (ex): the outer commissure of each eye fissure. Annotate the left first, then right. Verify correct placement in the frontal view.
   3. Glabella (g): most anterior midpoint on the fronto-orbital soft tissue contour. Identify the landmark placement on right and left profile views and then verify midline alignment in the frontal view.
2. Annotate the landmarks with the use of the Place landmarks option from the "Tools" menu of the "Landmarks" option in the top toolbar or from the Place landmarks option in the left-side toolbar (top arrow icon). Remember to annotate the landmarks in the same sequence every time for all the images that are going to be compared. To edit the location of a placed landmark select the Select landmarks option from the left-side toolbar (bottom arrow icon) and then the Place landmarks option which allows you to drag the selected landmark, by holding down the left key, to the correct position.
3. Save the landmarked image.

4. Selection of areas of interest in the baseline image and volume measurement

1. In the frontal view, select the Pick Multiple Points for Closed Loop from the left-side toolbar tool and place points along the perimeter of the region of interest.
2. After completing the loop, select Add to Area by the menu that appears after right-clicking on the screen.
3. Use the Paint Area Selection tool from the "Tools" menu of the "Area" option in the top toolbar to refine the selection by adding or erasing from the region of interest.
4. Perform the same procedure in the lateral and submental views to cover the region.
5. Once satisfied, use Use the Copy Selected Area option from the "Area" menu in the top toolbar to create a mask.
   NOTE: In this protocol, the region of interest was defined as that affected by orthognathic surgery and delineated by the lateral and medial canthi, glabella, and cervico-mental angle.

5. Registration of subsequent images, superimposition of area of interest, and volume measurement

1. Open both the baseline image and image that will be registered to the baseline axis grid (registration image).
2. Select the Show One Surface in Each of Two Viewports option in the top toolbar. 
3. Use the Spin tool (eye icon) in the left-side toolbar to manually rotate the image until its orientation closely matches that of the baseline image.
4. Once aligned visually, place the landmarks in the same sequence as described in step 3.
5. Select Register Surfaces from the "Surfaces" menu in the top toolbar and apply the following settings: move this surface "image" to fit this surface "baseline" and calculate alignment using "landmarks with corresponding names."
6. Switch to a single viewport  by selecting the "1 Viewport" option from the top toolbar and check the registration by performing Color Surface by Distance from the "Distance" menu  in the "Measure" menu options in the top toolbar with the following selections: color whole of, this surface "baseline" by the distance to this surface "image."
   NOTE: Unaffected or unaltered regions should show minimal color variation if registered well. Make sure to save the registered image.
7. Return to two viewports by selecting the "Show One Surface in Each of Two Viewports" option in the top toolbar, hide the registration image by clicking on the icon tab of the image, and use Project Selected Area from the "Area" menu on the top toolbar to highlight the corresponding region from baseline onto the second image.
8. Return to the two viewport view option and deselect the registration image icon so that it is not visible. From the "Area" menu select Hide to hide the selected area from the baseline image. Click on the second image and then select the Between Two Surfaces (different object) option from the "Volume" options under the "Measure" menu.
9. Return to a single viewport view and hide the registration image by clicking on its icon. Select the +/- Volumes option from the "Surface" menu and then measure the volume between the two selected areas by using the "Of Closed Surface" option from the "Volume" options in the "Measure" menu. The results can be found in the "Log" area at the bottom of the window.

This protocol describes in detail a standardized, reproducible approach for the 3D assessment of facial soft tissue changes. Stereophotogrammetry allows for the accurate quantification of facial swelling by capturing surface changes over time. Using a landmark-guided registration process, postoperative images are aligned to a stable baseline, enabling superimposition and comparison of areas affected by surgery. An area of interest is defined using reproducible anatomical boundaries, and the corresponding volume is calculated in cubic millimeters (cm3). As each subject serves as their own internal control, absolute volumes are not directly comparable between subjects in this analysis.

Figure 1: Demonstration of three-dimensional volumetric measurement. Demonstration of the use of this protocol for the quantitative assessment of the amount of facial swelling after orthognathic surgery. (A) Baseline image acquired one year after the surgery, when postoperative swelling was completely resolved. The image was registered to a 3D axis grid using the following anatomical landmarks: bilateral medial and lateral canthi, glabella. with definition of the ROI. (B) Postoperative image acquired 1 week after surgery, when most patients experience the most swelling. It was registered to the same axis grid, with projection of the baseline ROI onto the image. (C) Superimposition of baseline and postoperative images to confirm alignment accuracy. (D) Volumetric masks generated from the ROI at both time points, with resulting volumetric measurement in cubic millimeters (cm3). (E) Heat map comparing volume differences between the masks; blue and green indicate increased volume, while yellow and orange indicate decreased volume as compared to the baseline image. A color-coded number index is included as a reference. The exact volumetric difference between the two masks is also provided as an output. Please click here to view a larger version of this figure.

3D imaging has become the standard of care in the fields of dentistry and craniofacial surgery and offers quantitative soft tissue assessments of high precision¹¹. The reproducibility and accuracy of the placement of anatomical landmarks on 3D images have also been validated in previous studies¹⁹. Therefore, stereophotogrammetry or 3D facial soft tissue images enables the acquisition of precise anthropometric measurements and the objective quantification of soft tissue changes through longitudinal image comparisons.

Prior literature on facial soft tissue assessment often describes methods that require extensive training in specialized image analysis software, the combination of different imaging modalities and types of software, or the annotation of multiple landmarks, which can necessitate substantial time and effort from users^{11,13,14,20,21,22,23,24,25,26}. An area-based superimposition is also offered as an option by different software and has been implemented in previous studies²⁷. This is an alternative superimposition tool that could be used instead of the manual landmark annotation. However, based on the practical experience of the authors after having tested both methods as part of a retrospective cohort study, landmark-based superimposition was an easier and more accurate option.

The main reason was that the area-based superimposition was more sensitive to small differences between the superimposed images, like, for example, the presence of even a single hair on the skin surface or the precise orientation of the head. This is an inherent problem in retrospective studies that could have been avoided with the use of a more standardized image acquisition protocol, as advised below. Nevertheless, previous publications using landmark-based superimposition do not include stepwise guidance of the process so that it can be easily replicated by other clinicians and/or researchers. We present here a standardized framework for the quantification of volumetric changes, which is already integrated into existing imaging software. This is a straightforward approach to volumetric soft tissue evaluations that does not require expertise in the field of image analysis to be completed.

Some of the technical challenges that are often encountered during the superimposition of facial surface images are related to the presence of facial hair. Depending on its length and thickness, facial hair can significantly compromise the accuracy of the scans because the skin surface is obscured. Consequently, the comparison between scans, which is a high-sensitivity, quantitative assessment, is going to be unreliable. Therefore, the existence of very minimal or, ideally, no facial hair is strongly recommended prior to the acquisition of the 3D facial scans. The subject's hair should also be placed away from the face, allowing for a clear view of the face, forehead, and ears.

In addition, patients with oral breathing patterns or lip incompetency due to underlying skeletal discrepancies often tend to posture with their lips apart. This can become a problem during the comparison of serial scans, especially when the position of the lips is not consistent. The position of the lips should be reproducible when two or more scans taken at different timepoints will be included in comparative volumetric evaluations. Finally, facial expressions should also be avoided during the acquisition of the scans for consistency and reproducibility reasons. It is recommended that all images be taken with the patient maintaining a neutral facial expression, with their lips closed, if possible, and keeping their teeth in contact in their habitual bite relationships. This position is considered the easiest to replicate over time.

Moreover, it is also important to instruct the subjects accordingly, so that they know exactly what they need to do and what the process of capturing the images includes. For the same reason, the personnel involved in the acquisition of the images should be trained and calibrated with each other, when more than one person is involved. This can significantly improve the quality of the scans and, subsequently, the accuracy of the quantitative evaluations.

In summary, this protocol provides a practical guide for future clinical assessments and research studies seeking to evaluate the impact of different types of therapeutic interventions on facial esthetics, as well as the progression of healing processes, in the case of longitudinal assessments. This type of assessment could be very informative in various fields of dentistry, and particularly in the fields of fixed and removable prosthodontics, orthodontics, and orthognathic surgery^{28,29,30,31,32,33,34}. Research studies focusing on soft-tissue response to dental and/or skeletal tissue interventions would provide valuable information about the effect of these various treatment modalities on the facial soft tissue contour, which is an essential component of the overall treatment result. In addition, longitudinal assessments of the soft tissue changes related to aging can be equally informative about the long-term effect that different types of treatment could have on the soft tissue profile. This information could be used in the future as part of customized treatment result simulation approaches, which could be an essential part of treatment planning and patient consultations.