fastMRI

Artifacts Classification

Hi, I would like to know if there are any studies related to the classification of artifacts (also presence / absence) of the knee dataset.
Furthermore, I’d like to know where to find the MRI information for each slices (t1-weighted / t2-weighted, gradient, etc.) of the knee dataset. In the “ismrmrd_header” I find only magnet informations and slices thickness.

Thank you so much.

Hello @SyMon_92, welcome to fastMRI!

I don’t know of studies related to classification of artifacts - are you referring to standard imaging artifacts or deep learning reconstruction artifacts? We discuss deep learning reconstruction artifacts in our recent paper reporting the results of the 2020 fastMRI challenge:

The main way I look at the sequence is usually with

hf.attrs['acquisition']

Which, for example, might output CORPD_FS for a coronal proton density scan with fat saturation.

Hi @mmuckley, thank you very much for the quick reply!

Yes, I’m referring to standard imaging artifacts (e.g. metal prosthesis, motion, etc.).

Thanks for the hf.attrs[‘acquisition’] command, it works great.
For the raw data of knee dataset, is there any command that for each slice, shows the specific position in the coronal axis? For example, I would need to select all MRIs with the same perspective for each scan.

Thank you

I don’t know of any studies of artifacts in the fastMRI data, but there could be something in the literature.

Hmm, I’m not sure I have a solution for the positioning. You could try to inquire the people at NYU for that information.

fastmri@med.nyu.edu