Request a dataset format for obtaining MRI embeddings

[zzxj303]

Dear workers：
First of all, I am grateful for your work and great contribution to medical research.I'm working on (img_net="SwinUNETR"
img_mode=2 # loads the imgnet, generates embeddings out of the MRIs input to the network, and saves them.). We want to use our own dataset, but we don't have a standard CSV format and TOML file. Can you please provide me with a complete format for reference regarding image embedding. Looking forward to hearing from you.

[xf3227]

Dear @zzxj303,

Thank you for reaching out!

Since the number of scans available for each individual may vary, we did not hardcode the entries in the configuration TOML file. Instead, these image entries are dynamically created during the data-loading process. You can find the implementation details at line 139 in dataset_csv.py.

Please let us know if you have any further questions during implementation.