Variation of the face in rest using 3D stereophotogrammetry

Abstract

To evaluate treatment outcomes following oral and maxillofacial surgery, pre- and post-treatment three-dimensional (3D) photographs of the patient’s face can assessed, but this procedure is accurate only if the face is captured with the same facial expression every time. The purpose of this prospective study was to determine variations in the face at rest; 100 3D photographs of the same individual were acquired at different times. Initially, 50 3D photographs were obtained; 25 using a wax bite to ensure similar occlusion between subsequent photographs and 25 without wax bite. This procedure was repeated 6 weeks later. Variation of the face at rest was computed. The influence of time and wax bite was investigated. Different anatomical regions were investigated separately. A mean variation of 0.25 mm (0.21–0.27 mm) was found (standard deviation 0.157 mm). No large differences were found between different time points or use of wax bite. Regarding separate anatomical regions, there were small variations in the nose and forehead regions; the largest variations were found in the mouth and eyes. This study showed small overall variation within the face at rest. In conclusion, different 3D photographs can be reproduced accurately and used in a clinical setting for treatment follow-up and evaluation.

Introduction

To evaluate the results of surgical interventions in oral and maxillofacial surgery, preoperative and postoperative three-dimensional (3D) photographs of the patient’s face can be registered using surface based registration . 3D photographs of the patient are acquired at different moments in time and compared after aligning them using a surface based registration method. After registration of the facial surfaces the differences can be visualized using a colour scale or distance map. In this way, results of a surgical intervention can be evaluated quickly, quantitatively and objectively. A 3D photograph results in a static picture of the patient, therefore the accuracy of the evaluation depends on the ability to capture the patients face in rest reproducibly on multiple occasions. The purpose of this study was to determine the variation of the face at rest using 3D photographs.

Materials and methods

In this prospective study, 100 3D photographs of the same volunteer were acquired. During acquisition, special attention was addressed to positioning the volunteer and relaxing the facial musculature. The volunteer was placed in a natural head position and was asked to bite in maximum intercuspidation, swallow, relax his lips and keep both eyes open. All 3D photographs were taken by a trained photographer using a five-point 3D sterophotogrammetrical camera setup (3dMDCranial™ System, 3dMD LLC, Atlanta, USA). At the first time point, 50 3D photographs were acquired. In between, subsequent 3D photographs a rest moments of 1 min were incorporated. To investigate the influence of the position of the mandible, the first 25 3D photographs were acquired without a wax bite (group A1) and the subsequent 25 3D photographs with a wax bite in place (group A2). The wax bite used in this study was acquired in maximal occlusion.

To evaluate the influence of time, the procedure was repeated 6 weeks later, resulting in another 50 3D photographs, 25 without a wax bite (group B1) and 25 with a wax bite in place (group B2). In this way, four separate groups were aqcuired ( Fig. 1 ).

Fig. 1
100 3D photographs of the same individual were acquired. At the first time point, 50 photographs were obtained. They were divided into two groups, 25 normal 3D photographs and 25 3D photographs in which a wax bite was used to assure similar occlusion. This procedure was repeated 6 weeks later, resulting in four different groups.

For every group, one 3D photograph was selected randomly to be the reference photograph. All other 3D photographs in the group were registered with this reference 3D photograph using surface based registration. Surface based registration of the different 3D photographs was performed using Maxilim v.2.3.0 ® (Medicim NV, Mechelen, Belgium) software. The registration procedure was repeated for all four groups. The 3D photographs of time point 2 (groups B1 and B2) were registered with the reference 3D photograph of time point 1 (groups A1 and A2) so investigating the influence of time. The groups with a wax bite during acquisition (groups A2 and B2) were registered and compared with the 3D photographs of the group without a wax bite (groups A1 and B1).

To investigate the regions of variation within the face, the 3D photographs were classified into several anatomical facial regions (forehead, eyebrows, eyes, nose, cheeks, mouth and chin) ( Fig. 2 ). The variation was computed for each region.

Fig. 2
The face was divided into several anatomical regions: forehead, eyebrows, eyes, nose, cheeks, mouth and chin.

To compute the variation between all registered 3D photographs in a group Matlab was used. The reference 3D photographs were divided into ±20.000 points. From each of these points the closest distance to all the other 3D photographs was computed. From these calculations, the variation between the different registered 3D photographs could be computed and statistical analysis could be performed. The root-mean-square (RMS) error, standard deviation and 90th and 95th percentile were computed for the 3D photographs. To visualize the RMS error and the distribution of the variance, a cumulative distribution was plotted. For the different anatomical regions, the RMS error, standard deviation and 90th and 95th percentile of the registration error were calculated. To illustrate the variation of all separate regions, a cumulative distribution plot was generated.

Materials and methods

In this prospective study, 100 3D photographs of the same volunteer were acquired. During acquisition, special attention was addressed to positioning the volunteer and relaxing the facial musculature. The volunteer was placed in a natural head position and was asked to bite in maximum intercuspidation, swallow, relax his lips and keep both eyes open. All 3D photographs were taken by a trained photographer using a five-point 3D sterophotogrammetrical camera setup (3dMDCranial™ System, 3dMD LLC, Atlanta, USA). At the first time point, 50 3D photographs were acquired. In between, subsequent 3D photographs a rest moments of 1 min were incorporated. To investigate the influence of the position of the mandible, the first 25 3D photographs were acquired without a wax bite (group A1) and the subsequent 25 3D photographs with a wax bite in place (group A2). The wax bite used in this study was acquired in maximal occlusion.

To evaluate the influence of time, the procedure was repeated 6 weeks later, resulting in another 50 3D photographs, 25 without a wax bite (group B1) and 25 with a wax bite in place (group B2). In this way, four separate groups were aqcuired ( Fig. 1 ).

Fig. 1
100 3D photographs of the same individual were acquired. At the first time point, 50 photographs were obtained. They were divided into two groups, 25 normal 3D photographs and 25 3D photographs in which a wax bite was used to assure similar occlusion. This procedure was repeated 6 weeks later, resulting in four different groups.

For every group, one 3D photograph was selected randomly to be the reference photograph. All other 3D photographs in the group were registered with this reference 3D photograph using surface based registration. Surface based registration of the different 3D photographs was performed using Maxilim v.2.3.0 ® (Medicim NV, Mechelen, Belgium) software. The registration procedure was repeated for all four groups. The 3D photographs of time point 2 (groups B1 and B2) were registered with the reference 3D photograph of time point 1 (groups A1 and A2) so investigating the influence of time. The groups with a wax bite during acquisition (groups A2 and B2) were registered and compared with the 3D photographs of the group without a wax bite (groups A1 and B1).

To investigate the regions of variation within the face, the 3D photographs were classified into several anatomical facial regions (forehead, eyebrows, eyes, nose, cheeks, mouth and chin) ( Fig. 2 ). The variation was computed for each region.

Fig. 2
The face was divided into several anatomical regions: forehead, eyebrows, eyes, nose, cheeks, mouth and chin.

To compute the variation between all registered 3D photographs in a group Matlab was used. The reference 3D photographs were divided into ±20.000 points. From each of these points the closest distance to all the other 3D photographs was computed. From these calculations, the variation between the different registered 3D photographs could be computed and statistical analysis could be performed. The root-mean-square (RMS) error, standard deviation and 90th and 95th percentile were computed for the 3D photographs. To visualize the RMS error and the distribution of the variance, a cumulative distribution was plotted. For the different anatomical regions, the RMS error, standard deviation and 90th and 95th percentile of the registration error were calculated. To illustrate the variation of all separate regions, a cumulative distribution plot was generated.

Results

The results for the different groups are illustrated in Table 1 . The mean error (RMS) of all four groups ranged from 0.21 mm to 0.27 mm. The standard deviation ranged from 0.20 mm to 0.26 mm. For the registration between time point 1 and time point 2 the RMS error increased to 0.36 mm. For the difference between the use of a wax bite or no wax bite, the RMS error was 0.35 mm.

Table 1
The RMS error, standard deviation and 90th and 95th percentile were computed for all 4 separate groups. these results were also calculated to investigate the difference between time point 1 and time point 2 and to investigate the use of a wax bite during acquistion of the 3D photographs.
RMS error Standard deviation 90th percentile 95th percentile
Time point 1
Without wax bite (A1) 0.2365 0.2487 0.5280 0.696
With wax bite (A2) 0.2148 0.2003 0.4579 0.5904
Time point 2
Without wax bite (B1) 0.2689 0.2599 0.5880 0.7735
With wax bite (B2) 0.2663 0.2475 0.5988 0.7529
Time point 1 vs. Time point 2 0.3632 0.3431 0.8058 1.0232
Wax bite vs. no wax bite 0.3498 0.4500 0.7219 1.0302

The 90% error was below 0.60 mm for all four groups and the 95% error was below 0.78 mm. The cumulative distribution ( Fig. 3 ), illustrated that a large amount of error was situated between 0.0 and 0.15 mm. The lines on the x -axis illustrate the 0.5 mm and 1.0 mm margin ( Fig. 3 ). From this plot it was clear that for 85–92% of all points the error was smaller than 0.5 mm. For an error of 1.0 mm this percentage is 98–100%.

Fig. 3
A cumulative distribution plot was generated to illustrate the variance of the separate groups.

The results for the different anatomical regions are illustrated in Table 2 . The RMS error ranged from 0.16 mm to 0.37 mm. The 90th percentile error ranged from 0.33 mm to 0.78 mm and the 95th percentile error ranged from 0.40 mm to 1.01 mm. The cumulative distribution plot depicting the variance in all anatomical regions is illustrated in Fig. 4 .

Tags:
Jan 27, 2018 | Posted by in Oral and Maxillofacial Surgery | Comments Off on Variation of the face in rest using 3D stereophotogrammetry

VIDEdental - Online dental courses
Get VIDEdental app for watching clinical videos