RandDeformGrid

[mariakawula]

Hi :)
My problem
I want to apply the following compose transform to my 3D data (mhd format):
Rotation + translation -> deformation field -> zoom/crop.
As long as I'm leaving the RandDeformGrid out it works but including it results in an error.
My question
Can I (and if so, how) include this deformation into my Compose transform?

        degree2euler = lambda d: d / 180.0 * np.pi
        train_im_trans = Compose( 
                [
                    LoadImage(image_only=True),
                    AddChannel(),
                    RandAffine(prob=1, rotate_range=(degree2euler(80), degree2euler(5), degree2euler(5)),
                               translate_range=(10, 10, 10), as_tensor_output=True),
                    RandDeformGrid(spacing=[1.0, 1.0, 1.0], magnitude_range=(5, 8), as_tensor_output=True),
                    RandZoom(prob=1, min_zoom=0.9, max_zoom=1.1, padding_mode='edge'),
                    CenterSpatialCrop((128, 128, 128)),
                    ToTensor(),
                ]
            )
        
            check_ds = ArrayDataset(images_my, train_im_trans, segs_my, train_seg_trans)
            check_loader = DataLoader(check_ds, batch_size=10, num_workers=2, pin_memory=torch.cuda.is_available())
            im, seg = monai.utils.misc.first(check_loader)

              File "/opt/.pycharm_helpers/pydev/pydevd.py", line 1477, in _exec
            pydev_imports.execfile(file, globals, locals)  # execute the script
          File "/opt/.pycharm_helpers/pydev/_pydev_imps/_pydev_execfile.py", line 18, in execfile
            exec(compile(contents+"\n", file, 'exec'), glob, loc)
          File "/data/main.py", line 176, in <module>
            main(tempdir)
          File "/data/main.py", line 67, in main
            im, seg = monai.utils.misc.first(check_loader)
          File "/opt/monai/monai/utils/misc.py", line 66, in first
            for i in iterable:
          File "/opt/conda/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 435, in __next__
            data = self._next_data()
          File "/opt/conda/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 1085, in _next_data
            return self._process_data(data)
          File "/opt/conda/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 1111, in _process_data
            data.reraise()
          File "/opt/conda/lib/python3.6/site-packages/torch/_utils.py", line 428, in reraise
            raise self.exc_type(msg)
        RuntimeError: Caught RuntimeError in DataLoader worker process 0.
        Original Traceback (most recent call last):
          File "/opt/monai/monai/transforms/utils.py", line 389, in apply_transform
            return transform(data)
          File "/opt/monai/monai/transforms/spatial/array.py", line 1077, in __call__
            self.spacing = fall_back_tuple(self.spacing, (1.0,) * len(spatial_size))
          File "/opt/monai/monai/utils/misc.py", line 167, in fall_back_tuple
            user = ensure_tuple_rep(user_provided, ndim)
          File "/opt/monai/monai/utils/misc.py", line 126, in ensure_tuple_rep
            raise ValueError(f"Sequence must have length {dim}, got {len(tup)}.")
        ValueError: Sequence must have length 1, got 3.
        The above exception was the direct cause of the following exception:
        Traceback (most recent call last):
          File "/opt/monai/monai/transforms/utils.py", line 389, in apply_transform
            return transform(data)
          File "/opt/monai/monai/transforms/compose.py", line 236, in __call__
            input_ = apply_transform(_transform, input_)
          File "/opt/monai/monai/transforms/utils.py", line 391, in apply_transform
            raise RuntimeError(f"applying transform {transform}") from e
        RuntimeError: applying transform <monai.transforms.spatial.array.RandDeformGrid object at 0x7f67f7466978>
        The above exception was the direct cause of the following exception:
        Traceback (most recent call last):
          File "/opt/conda/lib/python3.6/site-packages/torch/utils/data/_utils/worker.py", line 195, in _worker_loop
            data = fetcher.fetch(index)
          File "/opt/conda/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py", line 44, in fetch
            data = [self.dataset[idx] for idx in possibly_batched_index]
          File "/opt/conda/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py", line 44, in <listcomp>
            data = [self.dataset[idx] for idx in possibly_batched_index]
          File "/opt/monai/monai/data/dataset.py", line 914, in __getitem__
            return self.dataset[index]
          File "/opt/monai/monai/data/dataset.py", line 811, in __getitem__
            data.extend(to_list(dataset[index]))
          File "/opt/monai/monai/data/dataset.py", line 69, in __getitem__
            data = apply_transform(self.transform, data)
          File "/opt/monai/monai/transforms/utils.py", line 391, in apply_transform
            raise RuntimeError(f"applying transform {transform}") from e
        RuntimeError: applying transform <monai.transforms.compose.Compose object at 0x7f67b0a87e80>
        python-BaseException

[rijobro]

Hi, I don't think RandDeformGrid does what you're expecting it to do. For RandDeformGrid you pass spatial dimensions as argument, and it will return you a random control grid of this size.

I suspect you want to apply random non-rigid deformations? In which case you would want to use Rand3DElastic (or Rand2DElastic for 2D data).

Note that these elastic classes can also apply random rotations, translations, shears and scaling, so perhaps you can consolidate all your random transformations, and use just Rand3DElastic.

[mariakawula]

It solved my problem :D
Thank you soooooo much!

[ankitabuntolia]

can we read .mhd image file in Monai? I always knew that only nifti is supported but in your discussion, you mentioned using .mhd files so I am interested to know how can I read the .mhd file in Monai?

[Nic-Ma]

Hi @ankitabuntolia ,

MONAI can read most of medical image formats. Please ensure ITK is installed and have a try with MONAI LoadImage.

Thanks.