How to get data about modality type

[jakubMitura14]

Hello I had defined dataset and loader like below

check_ds = Dataset(data=data_dicts, transform=val_transforms)
check_loader = DataLoader(check_ds, batch_size=1)

Now I would like to obtain the information what modality the image is - CT/MRI/PET, and if possible for CT scan is it for hard or soft tissue image file so :

for dat in check_loader:
     print(dat.getModalityType())

Thanks for help !!

[Nic-Ma]

Hi @jakubMitura14 ,

Thanks for your interest here.
Besides raw image array, the LoadImaged transform also loads the meta data of NIfTI header or DICOM header, maybe you can find some information in the meta data.
@wyli @ericspod Do you have any other ideas or experience?

Thanks in advance.

[jakubMitura14]

Thanks ! - and how to get access to this metadata :)?

[Nic-Ma]

Please check this doc-string:
https://github.com/Project-MONAI/MONAI/blob/dev/monai/transforms/io/dictionary.py#L88
For example, if your image key is "image", metadata key will be "image_meta_dict".

Thanks.

[jakubMitura14]

maybe I am blind but after

    print(dat['image_meta_dict'])

I get the info as below and nothing seems to be related to modality type?

{'sizeof_hdr': tensor([348], dtype=torch.int32), 'extents': tensor([0], dtype=torch.int32), 'session_error': tensor([0], dtype=torch.int16), 'dim_info': tensor([0], dtype=torch.uint8), 'dim': tensor([[  3, 512, 512, 247,   1,   1,   1,   1]], dtype=torch.int16), 'intent_p1': tensor([0.]), 'intent_p2': tensor([0.]), 'intent_p3': tensor([0.]), 'intent_code': tensor([0], dtype=torch.int16), 'datatype': tensor([4], dtype=torch.int16), 'bitpix': tensor([16], dtype=torch.int16), 'slice_start': tensor([0], dtype=torch.int16), 'pixdim': tensor([[-1.0000e+00,  9.7656e-01,  9.7656e-01,  3.2700e+00,  0.0000e+00,
          1.0000e+00,  1.0000e+00,  4.3047e+04]]), 'vox_offset': tensor([0.]), 'scl_slope': tensor([nan]), 'scl_inter': tensor([nan]), 'slice_end': tensor([0], dtype=torch.int16), 'slice_code': tensor([0], dtype=torch.uint8), 'xyzt_units': tensor([10], dtype=torch.uint8), 'cal_max': tensor([0.]), 'cal_min': tensor([0.]), 'slice_duration': tensor([0.]), 'toffset': tensor([0.]), 'glmax': tensor([255], dtype=torch.int32), 'glmin': tensor([0], dtype=torch.int32), 'qform_code': tensor([1], dtype=torch.int16), 'sform_code': tensor([1], dtype=torch.int16), 'quatern_b': tensor([0.]), 'quatern_c': tensor([1.]), 'quatern_d': tensor([0.]), 'qoffset_x': tensor([250.]), 'qoffset_y': tensor([-249.0232]), 'qoffset_z': tensor([-978.]), 'srow_x': tensor([[ -0.9766,   0.0000,   0.0000, 250.0000]]), 'srow_y': tensor([[   0.0000,    0.9766,    0.0000, -249.0232]]), 'srow_z': tensor([[   0.0000,    0.0000,    3.2700, -978.0000]]), 'affine': tensor([[[   1.0000,    0.0000,    0.0000, -249.0000],
         [   0.0000,    1.0000,    0.0000, -249.0232],
         [   0.0000,    0.0000,    1.0000, -978.0000],
         [   0.0000,    0.0000,    0.0000,    1.0000]]]), 'original_affine': tensor([[[-9.7656e-01,  0.0000e+00,  0.0000e+00,  2.5000e+02],
         [ 0.0000e+00,  9.7656e-01,  0.0000e+00, -2.4902e+02],
         [ 0.0000e+00,  0.0000e+00,  3.2700e+00, -9.7800e+02],
         [ 0.0000e+00,  0.0000e+00,  0.0000e+00,  1.0000e+00]]],
       dtype=torch.float64), 'as_closest_canonical': tensor([False]), 'spatial_shape': tensor([[512, 512, 247]], dtype=torch.int16), 'original_channel_dim': ['no_channel'], 'filename_or_obj': ['C:\\Users\\1\\Downloads\\volumes 0-49\\volumes 0-49\\volume-19.nii.gz']}

Thanks!

[Nic-Ma]

Hi @ericspod @wyli ,

Is it possible to get some modality info from NIfTI header?
Or maybe it's better to get this info from the metadata.json of MONAI model package when we added it.

Thanks in advance.

[ericspod]

The NifTI header doesn't include modality information explicitly: https://nipy.org/nibabel/nifti_images.html#the-nifti-header There are fields for text like descrip where this could be stored but it's non-standard and not reliable to try to inspect these fields for something saying CT or MR.

[jakubMitura14]

ok but it is in DICOM - does data from Dicom gets propagated into metadata?

[wyli]

it's possible with the latest 0.8.1 release candidates using reader="itkreader and series_meta=True:

from monai.transforms import LoadImaged

key = "img"
loader = LoadImaged(keys=key, reader="itkreader", series_meta=True)
img = loader({key: "/Users/Downloads/DICOM/case1"})
print("metadict:", img[key+"_meta_dict"])

pip install monai==0.8.1rc4