Additional information regarding the dataset

Dear organizers of the FeTA challenge 2021,

Thank you for sharing this amazing dataset!

A bit more information about the dataset would be very useful. - were all the MRI acquired as part of clinical routine or did the patients volunteer? - are you considering to make publicly available the pathologies for all the 'pathological' cases? It looks like the IRTK cases include pathologies other than open spina bifida. - for the open spina bifida cases, were they operated on? If yes, which volumes are before / after surgery? - are all the volumes from different patients? - regarding the 'Neurotypical' cases, did they have other (suspected or confirmed) abnormalities? - what is the MR acquisition center for each volume? - which MR sequence was used for each volume?

If you ethics approval allows you to share those details, they would be of great value to interpret the segmentation results obtained using this dataset.

Best wishes, Lucas

Dear Lucas

Thank you for the interest in our dataset!

• In most cases, the images were acquired as part of the clinical routine, but we also included a few cases from healthy controls who volunteered for other, ongoing fetal MRI studies (so not specifically for the FeTA)
• Unfortunately we cannot disclose more detailed clinical information for the FeTA dataset at the single case level to ensure complete and irreversible anonymization and to hinder re-identification.
• Since the dataset is primarily for research purposes along the lines of the FeTA challenge, releasing more detailed clinical history and patient characteristics was not considered relevant. However, since we are collecting data from many spina bifida and other pathologies as well as controls with normally appearing neurodevelopment, a collaboration outside the scope of the FeTA challenge is something that we can discuss and which would of course permit sharing such information.
• the neurotypical cases had suspected neuroabnormalities but were later negative in the radiological (MRI) reports, or had confirmed, non-neuronal abnormalities with no known effects on prenatal neurodevelopment.
• the MR images were acquired in the same center for all cases. Please note that the reconstruction method is different (there are two types)
• the acquisition details are the same as in the first FeTA paper (describing a previous version of the dataset) https://arxiv.org/ftp/arxiv/papers/2010/2010.15526.pdf

We understand that the interpretation of the segmentation results might benefit a bit by knowing more about the underlying pathology at the case level, however, it is not permitted to release detailed diagnostic information. To make sure we rank the algorithms in a possibly fair way, additional metrics according to age / pathological status will be evaluated.

Hope this helps. Best Regards, Andras