Accuracy of Using Automatic or Blended Techniques for Mandibular Segmentation

Objectives: 3D printing or additive manufacturing, a process where layers are placed one at a time to create objects, is poised to revolutionize dental prostheses creation. This preliminary research is to assess the accuracy of 3D printed objects manufactured using either an automatic or blended (automatic and manual) segmentation process.
Methods: Two dry skull mandibles were scanned using an iCAT FLX cone beam CT machine. Scanning was done in different voxel resolutions (0.30mm, 0.25mm, and 0.20mm) at 16cm x 8cm field of view. The scans were exported as DICOM files to an open source segmentation software (3DSlicer). Segmentation of the mandible 1 was done using automatic segmentation for the three different scan resolutions. Mandible 2 was segmented using the blended approach. A total of six different 3D models were printed in a Fortus 400mc 3D printer at a thickness of .005 inch using ABS M30i (thermoplastic material used in 3D printing). Linear measurements of the two mandibles and six 3D printed models were measured and analyzed.
Results: Results indicated that there was no significant difference between the dry skull mandibles and the models but the condylar borders were better depicted in the blended segmentation models.
Conclusions: Current research concluded that using a blended approach for segmentation is a better alternative to automatic segmentation especially if finer structures are to be depicted with accuracy but the use of either technique is sufficient if it involves larger linear structures. Further research using angular and textural analysis has to be conducted to confirm the validity and reliability of using both segmentation techniques for different conditions.