DICOM to BIDS conversion for Hyperfine (Portable MRI) dataset

Hello everyone,

I have Hyperfine data that I would like to convert into BIDS specification. I’m curious about the correct naming conventions and how to configure the config.json file appropriately. Each person has only anatomical data, which includes the following sequences:
localizer, T1_AXI, T1_COR, T1_SAG, T2_AXI, T2_COR, and T2_SAG.

Best,
Hengameh

Assuming these are T1- and T2-weighted images, you could indicate variations with acq-<label>:

sub-01_acq-axi_T1w.nii.gz
sub-01_acq-cor_T1w.nii.gz
sub-01_acq-sag_T1w.nii.gz
sub-01_acq-axi_T2w.nii.gz
sub-01_acq-cor_T2w.nii.gz
sub-01_acq-sag_T2w.nii.gz

I don’t know what you mean about a config.json. I assume you’re using some specific converter?

1 Like

I would also check the JSON images and make sure conversion used the latest stable release of dcm2niix (v1.0.20230411). While prior versions are likely to work, this was the first release where I had access to hyperfine sample data.

Hyperfine has a ground breaking product that can fill many niches. However, the signal to noise is not in the same league as conventional MRI scanners. The tools in our field were developed for and training on these conventional scans. You should have realistic expectations for the performance of spatial operations like normalization (where slice thickness matters) and segmentation. Also, be aware that while T2 times decrease with field strength, T1 increase with field strength. Therefore, while you can acquire a T1 with any scanner using short TEs and TRs, and T2s with long TEs and TRs, the tissue contrast will be very different for ultra low field scanners. Indeed, this is why T1 scans from 3T and 7T look so different (e.g. the time of flight artifacts are much more pronounced at 7T due to T1 effects).

2 Likes

I appreciate your guidance. Indeed, I referred to the configuration file.