IADR Abstract Archives
Three-Dimensional Markerless Time-Sequential Smile Analysis
Objectives: Patients with cleft lip tend to show dysmorphology and distorted facial motion during facial expressions even after revision surgeries, because of scars. Although a marker tracking system has been employed to analyze facial motions, the facial surface has been ignored, and locating markers is time-consuming. The aim of the present study was to develop a novel three-dimensional (3D) markerless time-sequential smile analysis with facial surfaces and to compare its accuracy with the conventional method.
Methods: One subject with normal occlusion was enrolled. Thirty-eight markers were pasted on the face and used as the gold standard. The subject was asked to perform a sequential task, involving rest (2 s), maximal smile (3 s), and rest (2 s). After several rehearsals, the task was recorded seven times with a temporal 3D image-capturing device (Temporal-3dMD Systems, 3DMD, Atlanta, GA, USA) at a speed of 10 frames/s. As the pasted markers on the face could be captured only in color, we obtained the gold standard data from the color data by identifying the markers in each frame. Further, for each task and frame, wire mesh fitting was performed by means of a homologous modeling method based only on the 3D surface shape data. The estimated marker and gold standard were compared to calculate the accuracy. To visualize the data, a spring model was employed.
Results: Twenty-six landmarks located in the nasolabial motions showed an error of less than 2 mm in the vertical and anteroposterior directions. The spring model was considered useful to visualize the results.
Conclusions: A novel 3D markerless time-sequential smile analysis was developed to analyze the 3D facial surface of the nasolabial region, which allows a quantitative evaluation of the facial displacement in patients with cleft lip.
Methods: One subject with normal occlusion was enrolled. Thirty-eight markers were pasted on the face and used as the gold standard. The subject was asked to perform a sequential task, involving rest (2 s), maximal smile (3 s), and rest (2 s). After several rehearsals, the task was recorded seven times with a temporal 3D image-capturing device (Temporal-3dMD Systems, 3DMD, Atlanta, GA, USA) at a speed of 10 frames/s. As the pasted markers on the face could be captured only in color, we obtained the gold standard data from the color data by identifying the markers in each frame. Further, for each task and frame, wire mesh fitting was performed by means of a homologous modeling method based only on the 3D surface shape data. The estimated marker and gold standard were compared to calculate the accuracy. To visualize the data, a spring model was employed.
Results: Twenty-six landmarks located in the nasolabial motions showed an error of less than 2 mm in the vertical and anteroposterior directions. The spring model was considered useful to visualize the results.
Conclusions: A novel 3D markerless time-sequential smile analysis was developed to analyze the 3D facial surface of the nasolabial region, which allows a quantitative evaluation of the facial displacement in patients with cleft lip.