Hello, It seems that MLCUBE requires me to set the path of the input data file and the output path of the prediction results. I'm wondering what these paths should be set to? And are the format requirements for outputting prediction results and submission to leaderboard the same? Best Regards

Created by Tianze Tang TianzeTang
@TianzeTang , Thank you for the notice! Please see [this thread](https://www.synapse.org/Synapse:syn53708249/discussion/threadId=11453&replyId=33153) for an updated compatibility test command for the Post-Op segmentation models.
@vchung I think there are something wrong with Compatibility test for Segmentation task I. The test program says the image size is [240,240,155] and output origin should be [0, -239, 0]. They are 2023 Task 1 data. I think the size should be [182,218,182] and origin should be [-90, 126, -72] this year.
Thank you very much!
@TianzeTang , If I understand your first question correctly, you will not need to provide the filepaths to the `--data_path` and `--output_path` parmeters of your `mlcube.py` script. MedPerf will be passing these arguments when running your model MLCube. As for your second question - Yes, please follow the same submission criteria from the validation phase, that is: **Segmentations** ```text output_path ??? BraTS-{dataset}-{case_id}-{timepoint}.nii.gz ??? BraTS-{dataset}-[case_id}-{timepoint}.nii.gz ??? ... ``` **Global Synthesis (Missing MRI)** ```text output_path ??? BraTS-GLI-{case_id}-{timepoint}-{modality}.nii.gz ??? BraTS-GLI-{case_id}-{timepoint}-{modality}.nii.gz ??? ... ``` **Local Synthesis (Inpainting)** ```text output_path ??? BraTS-GLI-{case_id}-{timepoint}-t1n-inference.nii.gz ??? BraTS-GLI-[case_id}-{timepoint}-t1n-inference.nii.gz ??? ... ``` **Pathology** ```text output_path ??? {filename}.csv ``` where `{filename}` can be whatever filename you choose. Please note that it is expected for the output folder to only contain a single (2-column) CSV file.
hi, which challenge are you participating?

Inquiry about MLCUBE path and output format page is loading…