Developing methods of matching 3D facial images with computed tomography data

Abstract

The need to introduce digital technologies in facial prosthetics is dictated by an increase the number of oncological disease of the middle zone of the face and defects in the middle zone of the face of various etiologies. This study aimed to develop a method for matching three-dimensional images of the face with CT data and confirm its effectiveness.
Materials and methods.
To develop a methodology for combining virtual three-dimensional images of the face with CT data, reference objects in the form of 4 balls with diameter 5 mm were used. Scan and CT data were combined in Exocad (Germany). For the accuracy of matching 3D-scans with CT data, we chose 4 points on the patient's face, where the manufactured balls were attached: on the nose, on the chin and in the cheek area on both sides. Then the CT was done and facial 3D-scans were taken using scanners: 3dMDFace System (3dMD, USA/UK), ObiScanner (Fifth Ingenium, Italy), Einstar (Shining 3D, China). The obtained data were combined in Exocad. The data obtained on the Planmeca ProMax 3D ProFace (Finland) were used as a reference model, since this setup allows to get both a CBVT and three-dimensional photograph in one scanning pass, the patient's position, facial expression and muscle arrangement are the same for both specifield images, which ensures their exact compatibility. To check the distances between points obtained from the reference model, we used the InVesalius 3 (Brazil), which creates a 3D visualization of images based on a sequence of 2D DICOM files.
Results.
The conducted researches confirm that the error of matching 3D-face scans with computed tomography data (the difference in distances between the selected bone and soft tissue points) using the ObiScanner compared to Planmeca ProMax 3D ProFace reference model is 0.09—0.16 mm. The error using 3dMDFace System scanner is 0.20—0.40 mm. The error in compiling three-dimensional data between each other using Einstar scanner is 0.35—0.60 mm. By combining 3D face scans with computed tomography data on reference objects, we observe a slight deviation from the reference model (0.09—0.16 mm), that confirms the accuracy of the method of combining data on reference objects.
Conclusion.
The obtained statistical data, the conclusion is combination of three-dimensional data of face scanners with CT data by objects is the most accurate technique, taking into account suitable source three-dimensional data of face scan.