Could open photogrammetry be an alternative to 3D Orthodontics?
SD, Master in Orthodontics
The effectiveness of scans by intraoral scanners is undeniable. These are the gold standard and reference of any professional who wishes to use the new 3D technologies for planning procedures involving orthodontics. However, while these technologies meet the needs of the market, they also stand out because of the high costs. Few professionals, taking into account the whole, have access to these means in a constant and unrestricted way.
The purpose of this material is to offer an affordable alternative to those who want to enter the world of 3D graphics, but do not have the means to enjoy powerful and expensive machines.
The objective is not to criticize the values, since, in the face of the investments and the results are quite coherent, in addition the world is very big and has place for all. What we are doing is just showing a way to achieve compatible results, with due limitations, on those produced by market references.
As discussed above, we have a problem which is the high cost of cutting-edge technologies. To create a workaround we need to look at the tools that key stakeholders have at their disposal.
Few offices have a 3D scanner, but everyone, in theory, has the possibility of generating a plaster model. Then the dental arches in plaster models can be stored and made available at the moment they need them, without the need for the constant presence of that patient.
Another feature that experts have is a digital camera, either the equipment itself or even a smartphone that performs this function well.
In addition, professionals will need a computer, be it a tower or even a notebook. The programs used in our tests are multiplatform because they run on Windows, MacOSX and Linux. In addition they are free, that is, a resource that is already available, just download them and in some cases compile them.
In summary, to carry out the experiment the following resources were necessary:
The elements that have given rise to this experiment are being studied by the authors since the year 2011.
In 2013, we agreed to use lines in homogeneous color structures to help the photogrammetry algorithm to create clouds of denser points and, therefore, models more coherent with the original part.
Source: Scanning faces into 3D from photos:
The methodology proved to be very effective when applied in alginate molds, resulting in ready-to-use models for the preparation of digital prostheses, which would later be printed and applied to the veterinary patient.
Source: The bionic animals - FUTUREMAG - ARTE (French-Germany): https://www.youtube.com/watch?v=-z1CIw6Z6tQ
The use of the technique in dentistry happens in 2016 with the case of Hanna the dog, a labrador who lost part of one of the incisors and had it recovered through virtual mirroring and 3D printing in chrome-cobalt. As you can see in the image above the photogrammetry was shown to be quite accurate, allowing the generation of sub-millimeter fittings.
Fonte: Adorable puppy left depressed and hungry can now eat again after a 3D printed tooth replaces one she broke while chewing: http://www.dailymail.co.uk/sciencetech/article-3678651/Animal-avengers-rescue-Adorable-puppy-eat-3D-printed-tooth-replaces-one-broke-chewing.html
Summarizing the background:
The next step would be to test on human dental arches.
Scanner: Sirona Orthophos XG 3D
Cone beam tomograph
Camera: Nikon D5200 macro lens 100mm circular flash sigma
Smartphone: iPhone 6
All photogrammetry work, tomography reconstruction, resizing and alignment of the meshes were performed in Blender, through the LiberTeeh3D addon that is under development.
This addon in turn groups a series of tools:
For the comparison between the meshes was used the CloudCompare software.
The photographs were taken by an individual who had never worked with photogrammetry. We asked him to photograph the arcades in a circle with two heights. Due to the lack of training, the number of photos ranged from 36 to 56 shots.
The lighting settings were under the responsibility of the individual who photographed the scene, in which case he chose to work the brightness and contrast so as to hide the scene (in black), even if the scene was darkened.
The low illumination apparently did not compromise the results, at least as far as digitizing the digital camera is concerned.
You can clearly see the quality of detail in the scanning performed by the 3D scanner. The CT scan also has a well structured mesh, but with less detail than the scanner. The photogrammetry performed from the digital camera was directly compatible with the reconstructed mesh from the tomography.
The discrepancy in the upper part of the meshes happened because of a change suffered in the scanning by 3D scanner. The purpose of the study lies in the region of the teeth, so the rest should be ignored.
There has been a problem in the generation of the mesh made by the reconstruction of the computed tomography, which is not common, since it is a mold created with only one type of material.
To compare the aligned meshes we chose to erase any region that was more (+), or less (-) 0.30 mm from the mesh created by the scanning by the 3D scanner.
As we can attest in the profiled graphs, the discrepancy between the meshes created by the CT-Scan and photogrammetry processes do not present great distortions when compared between them and do not present great difference when compared to the mesh generated by the 3D scanner, evidently tending to zero.
Here we see evidenced the problem in generating the mesh by the CT-Scan. Look that the incisors were structurally compromised.
Although the Smartphone vs. 3D Scanner comparison shows a greater distribution of distances, the difference between the meshes stayed within the established limits and reinforced the tendency to zero.
Apparently yes, it is possible to use open source photogrammetry tools in the field of 3D Orthodontics!
Obviously this is just a pilot experiment and we are working on one with a bigger n. We intend to address the concepts involved more broadly and at the same time solve the problems presented to us:
The first problem seems to be easy to solve, since, as mentioned above, such a discrepancy in reconstruction is not common.
The second problem may be related to the lack of more abundant lighting in the scene. As discussed in previous studies, this factor may compromise the reconstruction of a structure.
After the results, we formulated a ranking of the scans:
1st - 3D Scanner
2nd - Digital camera
3rd - CT-Scan
4th - Smartphone
This is the end of this first phase of testing. Other studies will be presented in the future, with more scans and certainly with a broader and more complete approach.
See you later!