Dcm2bids cannot differentiate between non-normalized MRI data and normalized MRI data

[uriartel]

I'm attempting to convert dicoms to niftis using the dcm2bids command.

In the config.json file, I added a custom label to ensure that only normalized data is being converted to niftis.

{
            "dataType": "anat",
            "modalityLabel": "T1w",
            "customLabels": "rec-normalized",
            "criteria": {
                "SeriesDescription": "T1w_MPR_vNav",
                "ImageType": [
                    "NORM"
                ]
            }
        },

If the subject only has one run of normalized data but multiple non-normalized T1 runs, the dcm2bids command will still convert all T1's into niftis regardless of the custom label. This issue occurs both for the T1's and the T2's.

Other than manually deleting all non-normalized data, is there way to exclude them in the config.json file to prevent this error from occuring?

Here is the original format, and I want to exclude the ORIGINAL, PRIMARY, M, and ND categories.

Thanks in advance!

{
            "dataType": "anat",
            "modalityLabel": "T1w",
            "criteria": {
                "SeriesDescription": "T1w_MPR_vNav",
                "ImageType": [
                    "ORIGINAL",
                    "PRIMARY",
                    "M",
                    "ND",
                    "NORM"
                ]
            }
        },

[dlevitas]

Hi @uriartel,

I've also run into protocols with non-normalized T1w images. For me, I use SideCarFilename instead of SeriesDescription, since both the regular T1w and non-nromalized T1w tend to have the same series description in the protocol. Here's an example from a portion of a configuration file (dcm2bids version 2.1.4):

{
         "dataType": "anat",
         "modalityLabel": "T1w",
         "criteria": {
            "SidecarFilename": "003*"
         }
      },

In this protocol example, a localizer was run first, followed by the T1w scan. So in terms of SideCarFilename the localizer was "001*", the non-normlized T1w was "002*", and the regular T1w was "003*". Since I didn't want to convert the localizer and non-normalized T1w, those were left out of the configuration file.

Hopefully this provides some assistance.

[IvanAlvarez]

Hi,

In case anyone else lands on this page looking for help, I found another solution for distinguishing between normalised and non-normalised data. The problem with @dlevitas solution is that it relies on the sidecar filenames being in a consistent and predictable order. In their example, 002 corresponding to the non-normalised data and 003 corresponding to the normalised data.

In SIEMENS sequences, the ImageType field contains the field "NORM" when the image is normalised. So, when I specify my .json configuration file, I use:

{
        "dataType": "func",
        "modalityLabel": "bold",
        "criteria": {
            "SeriesDescription": "<some_relevant_series_descriptor>",
            "ImageType": ["ORIGINAL", "PRIMARY", "M", "ND", "NORM", "MOSAIC"]
            },
}

Where the contents of ImageType matches the normalised images, but not the non-normalised images, which read:

"ImageType": ["ORIGINAL", "PRIMARY", "M", "ND", "MOSAIC"]

[arnaudbore]

Dear @IvanAlvarez.
Glad @dlevitas was able to help you with this.
Thank you @dlevitas