Incorporate virtual fit results into Session (#179)

[ebrahimebrahim]

Closes #179

Session.virtual_fit_results

Virtual fit results take the form of a dictionary mapping target IDs
to pairs (approval, transforms), where:

• approval is a boolean indicating whether the virtual fit for that target has been approved, and
• transforms is a list of transducer transforms resulting from the virtual fit for that target.

The idea is that the list of transforms would be ordered from best to worst, and should of course
contain at least one transform.

standardize_database

This is a utility script for developers to read in and write back out the dvc database. It is useful for standardizing the format of the example dvc data, and also for checking that the database mostly still works.

I have been using this on my own to help catch little discrepancies that pop up between the example data and the openlifu library -- I figured it's probably good to version control it here under a scripts/ folder since it needs to evolve alongside the library.

To use this script, install openlifu to a python environment and then run the script providing the database folder as an argument:

python standardize_database.py db_dvc/

Parallel SlicerOpenLIFU PR

(todo)

Remaining to-do

• Update DVC data
• Start work on a parallel PR on SlicerOpenLIFU that makes the virtual fitting placeholder be compatible with this breaking change
  • It is here: Save virtual fit results in scene and sessions p#179 SlicerOpenLIFU#179

[ebrahimebrahim]

@arhowe00 This PR is coming in parallel to OpenwaterHealth/SlicerOpenLIFU#179 which probably @sadhana-r will end up reviewing. But this one should be merged first and is completed, so you can start reviewing it. Besides the new virtual fit result format, I lumped in a script I've been using from time to time to make sure we haven't broken the database. LMK if you have any questions or if you want to have a quick meeting to clarify what's going on with virtual fitting and all that. TY!

[arhowe00]

Great job, I had just a few minor comments and questions. Also suggested adding checking the users table in your script.

[ebrahimebrahim]

(The rebase required merging the dvc data changes in main -- it was just the stuff you added in #214 -- should be good now)

[arhowe00]

Curious -- why did you make the 4th row of this matrix like this, just to test, or is this an extra scaling transformation across all dimensions? I assume you apply this like $matrix@(x,y,z,1)^T$. I take it that in this case, you would then divide everything by $0.05y + 1.4$ after the scale and translation (in this case, no translation)?

[ebrahimebrahim]

Correct, you would apply it like that. I think typically you would want to make the last row [0,0,0,1] which is probably what you were expecting. In this case I wasn't thinking anything fancy... I just want to stick any old 4x4 array in here and have tests that the array data is handled correctly -- not caring about the interpretation as an "affine transform".

[ebrahimebrahim]

Note to self: Do not merge yet please! I want to finish off OpenwaterHealth/SlicerOpenLIFU#179 first since this is a breaking change for SlicerOpenLIFU, and I don't want to cause trouble for other work you guys are doing on SlicerOpenLIFU

[ebrahimebrahim]

@arhowe00 I found a small issue and just pushed a fix, the commit message should explain it. Letting you know since you had already approved. Thanks!