Gen3 input data from MIDRC BIH - May 2025

[cgmeyer]

This is the first version of the MIDRC BIH input data: Modality/StudyDescription combinations with frequency counts and platform (the data resource(s) where the imaging study metadata originated) added.

[fedorov]

@cgmeyer I am guessing this one is superseded by #36 that I just merged - can you confirm?

[cgmeyer]

This one contains the imaging hub (MIDRC BIH) data: "in/BIH_StudyDescriptions_Gen3.tsv". The one you just merged is only for the MIDRC data commons "in/StudyDescriptions_Gen3.tsv". MIDRC central requested that we prioritize mapping the Duke CSpineSeg dataset with description "MRI CERVICAL SPINE WITHOUT CONTRAST".

We should meet to further discuss the format of the BIH_... input table, because I think there are outstanding questions about whether the format makes sense.

[fedorov]

Ah, right ... Initially I thought I missed/forgot to merge it, but I now I do remember there were some open questions...

[fedorov]

@cgmeyer it would be great if you could clarify what is the frequency number - studies? series? something else?

Specifically, I am confused why NLST-LSS, which shows up 3 times, has frequency 19569 for SEG modality

midrc_dicom_harmonization/in/BIH_StudyDescriptions_Gen3.tsv

Line 4 in 686b1c2

SEG NLST-ACRIN IDC 19569

but 6553 for the SR row

midrc_dicom_harmonization/in/BIH_StudyDescriptions_Gen3.tsv

Line 15 in 686b1c2

SR NLST-LSS IDC 6553

and only 761 for CT

midrc_dicom_harmonization/in/BIH_StudyDescriptions_Gen3.tsv

Line 64 in 686b1c2

CT NLST-LSS IDC 761

Those questions aside, I suggest we update this BIH list to have one row for each distinct value of StudyDescription, and contains aggregated distinct values for all modalities and all source repositories. Frequency I think ideally should contain the total number of distinct studies that contain it, but I am not sure if you have that information.