Datasets:

---

Forithmus

/

MR-RATE-coreg

like 4

Follow

forithmus 53

Thanks for releasing MR-RATE-coreg! While mirroring the full dataset I found that a set of studies in the backfilled batches (batch06, batch08, batch09) are broken at the source. There are two distinct defects, both confined to those three batches.

Summary

For comparison, native Forithmus/MR-RATE has 98,334 studies; Forithmus/MR-RATE-coreg resolves to 98,276 usable studies — exactly the 58 empty-zip studies short.

How to reproduce

Empty zips (each is 22 bytes = an empty-zip EOCD record, namelist() returns []):

import os, zipfile
from huggingface_hub import hf_hub_download
z = hf_hub_download("Forithmus/MR-RATE-coreg", repo_type="dataset",
                    filename="mri/batch06/ERDLB5A752_coreg.zip", local_dir="/tmp/probe")
print(os.path.getsize(z), zipfile.ZipFile(z).namelist())   # -> 22 []

Truncated inner files (zip opens fine, but one member fails to gunzip):

import gzip
gzip.open(".../mri/batch06/CMMEKOSEET/coreg_img/CMMEKOSEET_coreg_flair-raw-cor.nii.gz").read()
# -> EOFError: Compressed file ended before end-of-stream   (some are 0 bytes)

Re-downloading a fresh copy from the Hub does not fix either case, so the defect is in the uploaded files themselves, not in transfer.

Defect 1 — 58 empty *_coreg.zip archives (0 entries, 22 bytes)

mri/batch06/ERDLB5A752_coreg.zip
mri/batch06/GOQXIIQIXC_coreg.zip
mri/batch06/HAPTMFJ3BT_coreg.zip
mri/batch06/HYEXCSF65T_coreg.zip
mri/batch06/IAPMNPWLAH_coreg.zip
mri/batch06/IH4TT5C33U_coreg.zip
mri/batch06/IKXS6X5BE5_coreg.zip
mri/batch06/IQOCK3QR3B_coreg.zip
mri/batch06/IVX67QANQY_coreg.zip
mri/batch06/IW6P73PTUS_coreg.zip
mri/batch06/IXMLVYDE3F_coreg.zip
mri/batch06/IYBRFCVZD5_coreg.zip
mri/batch06/JG7TAECPQM_coreg.zip
mri/batch06/JGABP75LIO_coreg.zip
mri/batch06/JRLLRJ5LXT_coreg.zip
mri/batch06/JXQDBTEQMP_coreg.zip
mri/batch06/JZZ4WTWR6C_coreg.zip
mri/batch06/KBZ3F55DTN_coreg.zip
mri/batch06/KFREC5YISO_coreg.zip
mri/batch06/KIONBEY6YJ_coreg.zip
mri/batch06/KNJJP77GKN_coreg.zip
mri/batch06/KNWXP77STF_coreg.zip
mri/batch06/KOD3R5TPMX_coreg.zip
mri/batch06/KQMOHD5QXI_coreg.zip
mri/batch06/KR6OLDH6AI_coreg.zip
mri/batch06/L2QQCFKJQN_coreg.zip
mri/batch06/MTAPAQDT7Z_coreg.zip
mri/batch06/N4WNSNUWNF_coreg.zip
mri/batch06/O7JLORCD4J_coreg.zip
mri/batch06/OI7WORS5DA_coreg.zip
mri/batch06/ONKQ5UZPUI_coreg.zip
mri/batch08/IE5QYINJO7_coreg.zip
mri/batch08/INBG6HTJQR_coreg.zip
mri/batch08/ISVVASXBWV_coreg.zip
mri/batch08/IUKEG36T32_coreg.zip
mri/batch08/J5XJM5E4GT_coreg.zip
mri/batch08/JVBNUTZFKN_coreg.zip
mri/batch08/JWMMGWEYCU_coreg.zip
mri/batch08/L2ID6VDVML_coreg.zip
mri/batch08/LINK7HSQDJ_coreg.zip
mri/batch08/LNO7EQHB2G_coreg.zip
mri/batch08/LQDJAS7JNG_coreg.zip
mri/batch08/M3WM6K4RA5_coreg.zip
mri/batch08/N7OXUC6Z3O_coreg.zip
mri/batch08/O3LFSHSFG5_coreg.zip
mri/batch08/OGJVON6ZMK_coreg.zip
mri/batch08/PDHT7LSYOR_coreg.zip
mri/batch08/PMIO7UMWEW_coreg.zip
mri/batch08/PQZKC74I2M_coreg.zip
mri/batch08/Q26FEP53KA_coreg.zip
mri/batch08/Q5D5BX3EU3_coreg.zip
mri/batch08/SWWCXV2OI7_coreg.zip
mri/batch09/VBY7GKXLZ6_coreg.zip
mri/batch09/YJEOGMUA23_coreg.zip
mri/batch09/YSGFWBS57G_coreg.zip
mri/batch09/YVLCTISORO_coreg.zip
mri/batch09/ZHOETT6JDQ_coreg.zip
mri/batch09/ZISJORIWUW_coreg.zip

Defect 2 — 12 truncated/zero-byte .nii.gz inside otherwise-valid zips

Listed as zip → member:

mri/batch06/CMMEKOSEET_coreg.zip → CMMEKOSEET/coreg_img/CMMEKOSEET_coreg_flair-raw-cor.nii.gz
mri/batch06/ELUURDMH2C_coreg.zip → ELUURDMH2C/coreg_img/ELUURDMH2C_coreg_flair-raw-axi.nii.gz
mri/batch06/FLQ3HMOK4X_coreg.zip → FLQ3HMOK4X/coreg_img/FLQ3HMOK4X_coreg_t1w-raw-sag-2.nii.gz
mri/batch06/GABG4PMQGT_coreg.zip → GABG4PMQGT/coreg_img/GABG4PMQGT_coreg_flair-raw-axi.nii.gz
mri/batch06/GBUZWE6GMO_coreg.zip → GBUZWE6GMO/coreg_img/GBUZWE6GMO_coreg_flair-raw-axi.nii.gz
mri/batch06/HSNGHNXBUE_coreg.zip → HSNGHNXBUE/coreg_img/HSNGHNXBUE_coreg_flair-raw-axi.nii.gz
mri/batch06/KJHBESPG6X_coreg.zip → KJHBESPG6X/coreg_img/KJHBESPG6X_coreg_flair-raw-sag.nii.gz
mri/batch08/IB57IIJ2UU_coreg.zip → IB57IIJ2UU/coreg_img/IB57IIJ2UU_coreg_t1w-raw-axi.nii.gz
mri/batch08/LZEAHKRLPI_coreg.zip → LZEAHKRLPI/coreg_img/LZEAHKRLPI_coreg_t1w-raw-axi.nii.gz
mri/batch08/MYJ4K7RDNT_coreg.zip → MYJ4K7RDNT/coreg_img/MYJ4K7RDNT_coreg_t2w-raw-axi.nii.gz
mri/batch08/N7LRDJW66Q_coreg.zip → N7LRDJW66Q/coreg_img/N7LRDJW66Q_coreg_flair-raw-axi.nii.gz
mri/batch09/VPP2MLWLGF_coreg.zip → VPP2MLWLGF/coreg_img/VPP2MLWLGF_coreg_t2w-raw-axi.nii.gz

Request

Could you re-generate and re-upload the coreg archives for these studies in batch06/08/09? All of the affected study UIDs are present and intact in the native Forithmus/MR-RATE repo, so the source images exist — it looks like the coregistration/zip step failed silently for these during the backfill.

Happy to provide any additional diagnostics. Thanks!

Best,
Moritz