Feature/add 2024 isbi goat algorithms - #144
Conversation
|
🤖 Code Formatting Reminder Hello there! 👋 It looks like the code in this pull request might benefit from some formatting improvements. |
|
/format |
|
🤖 I will now format your code with black. Check the status here. |
|
One test failed: tests/utils/test_algorithm_config.py::test_integrity_challenge_manuscript. I added the challenge manuscript for ISBI on purpose. It is separate from the Miccai BraTS challenge paper, hence we need a second challenge paper for the year 2024. I am unsure how to handle this - Should we modify or ignore the test ? |
|
With the help of copilot, I created a second PR that addresses this problem. The suggestion is to group not only by year, but by challenge and year: |
| @@ -1,9 +1,15 @@ | |||
| constants: | |||
| input_name_schema: &input_name_schema "BraTS-GLI-{id:05d}-000" | |||
| challenge: &challenge "BraTS Generalizability Across Tumors (BraTS-GoAT)" | |||
There was a problem hiding this comment.
improve naming to distinguish if we want tos tick with single file (would not recommend)
There was a problem hiding this comment.
I'm not familiar with ISBI - is this a separate challenge? If yes, should we also have a separate metadata file? would also avoid having this multi-paper change
There was a problem hiding this comment.
seems better if we have a separate segmenter anyways
| segmenter = ISBIGoATSegmenter( | ||
| algorithm=ISBIGoATAlgorithms.BraTS24_ISBI_1, | ||
| cuda_devices="0" | ||
| ) |
|
|
||
| class GoATAlgorithms(Algorithms): | ||
| """Constants for the available missing mri algorithms.""" | ||
| """Constants for the available missing mri algorithms. |
There was a problem hiding this comment.
preexisting issue but fix here: missing mri algorithm
| requires_root: false | ||
| parameters_file: true | ||
| additional_files: | ||
| record_id: "00000000" |
There was a problem hiding this comment.
this seems wrong? add proper id or remove
| | Year | Rank | Author | Paper | CPU Support | Key Enum | | ||
| | ---- | ---- | ------------------------- | ------------- | --------------- | --------------------------------------------------------------------------------------------------------------------------- | | ||
| | 2024 | 1st | _André Ferreira, et al._ | https://arxiv.org/abs/2402.17317 | ❌ | [BraTS24_ISBI_1](https://brats.readthedocs.io/en/latest/utils/utils.html#brats.constants.ISBIGoATAlgorithms.BraTS24_ISBI_1) | | ||
| | 2024 | 2nd | _Anees Hashmi, et al._ | [PAPER_URL_2] | [CPU_SUPPORT_2] | [BraTS24_ISBI_2](https://brats.readthedocs.io/en/latest/utils/utils.html#brats.constants.ISBIGoATAlgorithms.BraTS24_ISBI_2) | |
There was a problem hiding this comment.
insert proper values for paper and cpu support
|
@MarcelRosier Hi Marcel! Thank you so much for taking the time to review my PR! :) I read your comments and they seem absolutely valid and I really appreciate your feedback. The thing is, I intentionally published this only as a draft, because it was in fact not yet ready for review, but you could not have known that! ISBI is in fact a different conference, but it hosts the same BraTS Challenge, that is why I implemented it as it is now. But I totally see your point of this being messy and I will look into your idea of using a new metadata file! Maybe we do not need to modify the test then, but I have to check what makes most sense here. I will continue working on this as soon as I have all relevant information from the developers, but thank you again for taking the time to review! :) |
2 participants
This feature implements the top-2 algorithms from the IEEE-ISBI BraTS GoAT challenge from the year 2024. The images have already been pushed to the brainles docker hub page. The algorithm from the first place is executable, while for the second place, we are currently missing the correct Zenodo ID to download weights etc.