Without data, nothing can be done. To get you started, we recommend you download the Example data for Ivadomed. This dataset is composed of 10 subjects from different imaging centers and includes original images in NIfTI format as well as manual segmentations and labels. The data are organized according to the BIDS convention, to be fully compatible with ivadomed loader:

└── dataset_description.json
└── participants.tsv
└── sub-01
    └── anat
        └── sub-siteX01_T1w_reg.nii.gz
        └── sub-siteX01_T1w_reg.json
        └── sub-siteX01_T2w_reg.nii.gz
        └── sub-siteX01_T2w_reg.json
        └── sub-siteX01_acq-MTon_MTS_reg.nii.gz
        └── sub-siteX01_acq-MTon_MTS_reg.json
        └── sub-siteX01_acq-MToff_MTS_reg.nii.gz
        └── sub-siteX01_acq-MToff_MTS_reg.json
        └── sub-siteX01_acq-T1w_MTS.nii.gz
        └── sub-siteX01_acq-T1w_MTS.json
        └── sub-siteX01_T2star_reg.nii.gz
        └── sub-siteX01_T2star_reg.json
└── derivatives
    └── labels
        └── sub-siteX01
            └── anat
                └── sub-siteX01_T1w_seg.nii.gz


participants.tsv should, at least, include a column participant_id, which is used when loading the dataset.