I am an orthodontist and I would like to seek assistance regarding my extensive study.

[dtpathos]

I am unfamiliar with the software industry and would like to measure the distance changes in pre-treatment and post-treatment CBCT images of my patients through heat maps.

I learned how to perform alignment through YouTube videos; however, the tutorials mostly focused on large objects or even wide terrain data.

What parameters should a heat map of approximately the size of an average human head have to provide the most accurate results for me?

[daniel]

Not sure to understand: by heat map, do you mean the color ramp used to represent the distance values? If yes, the steps/distances for the various colors should probably be based on their clinical significance, shouldn't they?

And if you are concerned by the accuracy of the result, then you should pay attention to how well both models are initially roughly registered (the closer the better), and then the major parameter is the overlap ratio (if the 2 objects are complete and don't have missing parts, then 100% should be the default. But if they have missing parts - at least one with respect to the other - or some particular parts went under important changes while some others didn't change - then the overlap ratio should represent the percentage of surface that was unchanged / not missing.

[dtpathos]

Not sure to understand: by heat map, do you mean the color ramp used to represent the distance values? If yes, the steps/distances for the various colors should probably be based on their clinical significance, shouldn't they?

And if you are concerned by the accuracy of the result, then you should pay attention to how well both models are initially roughly registered (the closer the better), and then the major parameter is the overlap ratio (if the 2 objects are complete and don't have missing parts, then 100% should be the default. But if they have missing parts - at least one with respect to the other - or some particular parts went under important changes while some others didn't change - then the overlap ratio should represent the percentage of surface that was unchanged / not missing.

In fact, as you said, I will overlap the two scan images from unchanging points. I want to calculate the amount of change in millimeters of the places I expect to change. What kind of adjustment should I make? Because the changes will be at very minimal values.

[daniel]

The use ICP, with the right overlap parameter. See https://www.cloudcompare.org/doc/wiki/i ... n_with_ICP and https://www.cloudcompare.org/doc/wiki/index.php/ICP

[dtpathos]

The use ICP, with the right overlap parameter. See https://www.cloudcompare.org/doc/wiki/i ... n_with_ICP and https://www.cloudcompare.org/doc/wiki/index.php/ICP

Thanksi, have a nice day

[ortho-hendrik]

I am also using CC to register dental models and perform a ICP on the vestibular surfaces that I extract from dental models using CloudCompare. (see Image)

I works really great for me, but maybe as an addition to this topic (and since I am also an orthodontist using CC ;) I have the following question: is it possible to restrict the ICP analysis to certain areas of the two compared models (mesh)? When extracting the vestibular surface of the compared dental models, there will always be a certain deviation between the border of the resulting models. Im thinking about an option to exclude a defined amount of the outer perimeter of the mesh from the analysis. I am trying to quantify the deviation between the flat vestibular surfaces of the models, so I need to get rid of the deviation in the ICP that is caused by the uneven extraction of the vestibular surfaces or the gum (which is why I cut out the surfaces).

I would really appreciate any suggestions on this one! Cheers.

[daniel]

While it would be easy to ignore the points with the largest distance and restart the ICP process (with the 'Edit > Scalar fields > Filter by value'), it's not as simply with meshes. Either you remove the parts you want to ignore manually (with the scissors tool or the Cross section tool). Or you try your luck with 'Edit > Scalar fields > Filter by value' in case your mesh has a high density. Last option is to convert the mesh to a cloud (Edit > Mesh > Sample points).