ValueError: y_pred and y should have same shapes

[HripsimeS]

Hello @yiheng-wang-nv @wyli @rijobro @Nic-Ma
I am trying to optimize the "Spleen 3D segmentation with MONAI" tutorial on my dataset. Training is going well and with 400 epochs I get 96% best score. But as soon as I am trying to evaluate my model, I am ending up with the issue that y_pred and y don't have same shapes. Can you please check my notebook and see what is wrong with y_pred and y shapes. Thanks a lot in advance!

https://github.com/HripsimeS/Computer-Vision/blob/main/3DUnet_testing.ipynb

[yiheng-wang-nv]

Hi @HripsimeS , sorry for late reply. I checked your code, Where you used Spacingd(keys=["image", "mask"], pixdim=(1.0, 1.0, 1.0), mode=("bilinear", "nearest")):

val_org_transforms = Compose(
    [
        LoadImaged(keys=["image", "mask"]),
        EnsureChannelFirstd(keys=["image", "mask"]),
        Orientationd(keys=["image"], axcodes="RAS"),
        Spacingd(keys=["image", "mask"], pixdim=(1.0, 1.0, 1.0), mode=("bilinear", "nearest")),
        ScaleIntensityRanged(keys=["image"], a_min=-120, a_max=360,b_min=0.0, b_max=1.0, clip=True,),
        CropForegroundd(keys=["image"], source_key="image"),
        
        EnsureTyped(keys=["image", "mask"]),
    ]
)

Can you please change it into: Spacingd(keys=["image"], pixdim=(1.0, 1.0, 1.0), mode=("bilinear", "nearest"))?
Mask should not be changed, since in this part you'd like to calculate the metric on original shapes.

[HripsimeS]

Hello @yiheng-wang-nv. In fact I had to remove the "mask" from Orientationd, Spacingd and CropForegroundd. And I received almost the same score as I had for training. Thank you very much for your help 🥇 👍

[ashiwalv]

In my case, Spacingd without "label" throw error! It requires both keys. How did you made it to work with only one key as "image"?

[Nic-Ma]

Hi @ashiwalv ,

Could you please help share your code? Then @yiheng-wang-nv may help verify your issue.

Thanks in advance.

[ashiwalv]

Problem

Transform
If I use Spacingd(keys=["image", "label"], pixdim=(1.5, 1.5, 1.5), mode=("bilinear", "nearest")) then after transforming output shape of image is 1x128x128x128 and for labels 3x128x128x128.

If I tried to remove label as key from val_tranform then its throwing me error.

Errors when using spacingd without "Label" as key

ValueError                                Traceback (most recent call last)
[<ipython-input-79-a7080035a8ed>](https://localhost:8080/#) in <module>()
     70       ScaleIntensityRanged(keys='image', a_min=a_min, a_max=a_max, b_min=0.0, b_max=1.0, clip=False),
     71       Orientationd(keys=["image", "label"], axcodes="RAS"),
---> 72       Spacingd(keys="image", pixdim=(1.5, 1.5, 1.5), mode=("bilinear", "nearest")),
     73       # CropForegroundd(keys=["image", "label"], source_key="label", select_fn = 6000),
     74       NormalizeIntensityd(keys="image", nonzero=True, channel_wise=True),

1 frames
[/usr/local/lib/python3.7/dist-packages/monai/utils/misc.py](https://localhost:8080/#) in ensure_tuple_rep(tup, dim)
    142         return tuple(tup)
    143 
--> 144     raise ValueError(f"Sequence must have length {dim}, got {len(tup)}.")
    145 
    146 

ValueError: Sequence must have length 1, got 2.

Dataset: Image 512x496x49 and Label 512x496x49, labels containing 4 unique values 0= background and other 1,2,3, are classes. Its multilabel-segmentation task

Here is my tranform:

train_transform = Compose(
    [
      LoadImaged(keys=["image", "label"]),
      EnsureChannelFirstd(keys="image"),
      CopyItemsd(keys="image", times=3),
      SpatialCropd(keys=['image', 'label'], roi_start=(0,0,0), roi_end=(256,248,49)),
      SpatialCropd(keys=['image_1', 'label_1'], roi_start=(0,248,0), roi_end=(256,496,49),allow_missing_keys=True),
      SpatialCropd(keys=['image_2', 'label_2'], roi_start=(256,0,0), roi_end=(512,248,49),allow_missing_keys=True),
      SpatialCropd(keys=['image_3', 'label_3'], roi_start=(256,248,0), roi_end=(512,496,49),allow_missing_keys=True),
      ConcatItemsd(['image_1', 'image_2', 'image_3'], name='image', dim=1, allow_missing_keys=True),
      DeleteItemsd(['image_1', 'image_2', 'image_3']),
      ConvertToMultiChannelBasedOnOCTClassesd(keys="label"),
      ScaleIntensityRanged(keys='image', a_min=a_min, a_max=a_max, b_min=0.0, b_max=1.0, clip=False),
      Orientationd(keys=["image", "label"], axcodes="RAS"),
      Spacingd(keys=["image", "label"], pixdim=(1.5, 1.5, 1.5), mode=("bilinear", "nearest")),
      RandFlipd(keys=["image", "label"], prob=0.5, spatial_axis=0),
      RandFlipd(keys=["image", "label"], prob=0.5, spatial_axis=1),
      RandFlipd(keys=["image", "label"], prob=0.5, spatial_axis=2),
      Rand3DElasticd(keys=["image", "label"], mode=("bilinear", "nearest"), prob=1.0, sigma_range=(5, 8), magnitude_range=(100, 200),
                    spatial_size=(128, 128, 128), translate_range=(-140, 140), rotate_range=(np.pi / 36, np.pi / 36, np.pi), scale_range=(0.15, 0.15, 0.15), padding_mode="border"),
      NormalizeIntensityd(keys="image", nonzero=True, channel_wise=True),
      EnsureTyped(keys=["image", "label"]),
    ]
)
val_transform = Compose(
    [
      LoadImaged(keys=["image", "label"]),
      EnsureChannelFirstd(keys="image"),
      ConvertToMultiChannelBasedOnOCTClassesd(keys="label"),
      ScaleIntensityRanged(keys='image', a_min=a_min, a_max=a_max, b_min=0.0, b_max=1.0, clip=False),
      Orientationd(keys=["image", "label"], axcodes="RAS"),
      Spacingd(keys="image", pixdim=(1.5, 1.5, 5.0), mode=("bilinear", "nearest")),
      NormalizeIntensityd(keys="image", nonzero=True, channel_wise=True),
      EnsureTyped(keys=["image", "label"]),
    ]
)

P.S.: With my current tranform (using spacingd with both image and label keys), my model (unet from monai, lr =10-4) is not learning anything. any feedback on that would be helpful!

model = UNet(
    spatial_dims=3,
    in_channels=1,
    out_channels=3,
    channels=(16, 32, 64, 128, 256),
    strides=(2, 2, 2, 2),
    num_res_units=2,
    norm=Norm.BATCH,
).to(device)

[ashiwalv]

Thank you for your reply! Here is where I am getting errors without using labels as key in validation transform. #596 (reply in thread) and thanks for suggestions about inference mode! *Best Regards,* Virendra Ashiwal, Linz, Austria

…

On Sun, 27 Mar 2022 at 07:57, Hripsime Snkhchyan ***@***.***> wrote: @ashiwalv <https://github.com/ashiwalv> below label is removed from Orientationd, Spacingd and CropForegroundd which will calculate the metric on original shapes. Like this you can text your model on dataset outside of your training model, in test dataset you need to have images with their masks. val_org_transforms = Compose( [ LoadImaged(keys=["image", "label"]), EnsureChannelFirstd(keys=["image", "label"]), Orientationd(keys=["image"], axcodes="RAS"), Spacingd(keys=["image"], pixdim=(1.5, 1.5, 2.0), mode="bilinear"), ScaleIntensityRanged(keys=["image"], a_min=-57, a_max=164,b_min=0.0, b_max=1.0, clip=True,), CropForegroundd(keys=["image"], source_key="image"), EnsureTyped(keys=["image", "label"]), ] ) But if you want to test your model on inference mode, it means just upload the image and receive a predicied mask, you can do this test_org_transforms = Compose( [ LoadImaged(keys="image"), EnsureChannelFirstd(keys="image"), Orientationd(keys=["image"], axcodes="RAS"), Spacingd(keys=["image"], pixdim=(1.5, 1.5, 2.0), mode="bilinear"), ScaleIntensityRanged(keys=["image"], a_min=-57, a_max=164,b_min=0.0, b_max=1.0, clip=True,), CropForegroundd(keys=["image"], source_key="image"), EnsureTyped(keys="image"), ] ) — Reply to this email directly, view it on GitHub <#596 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAY5AKVY7OXOHZBMI5CMD3TVCABFLANCNFSM5PXKHD5Q> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[yiheng-wang-nv]

Hi @ashiwalv ,

Spacingd(keys="image", pixdim=(1.5, 1.5, 1.5), mode=("bilinear", "nearest"))

your keys is a single str, but mode is a tuple, I think your mode may be replaced with: mode="bilinear"

[MOMOANNIE]

Hello @yiheng-wang-nv @wyli @rijobro @Nic-Ma, because my label is very small, when I use the CropForegroundd function to crop the image with the label as the source_key, the shapes of y and y_pred are inconsistent. How should I solve this problem?

 elif dataset == 'flare':
        train_transforms = Compose(
            [
                LoadImaged(keys=["image", "label"]),
                AddChanneld(keys=["image", "label"]),
                Spacingd(keys=["image", "label"], pixdim=(
                    1.0, 1.0, 1.2), mode=("bilinear", "nearest")),
                # ResizeWithPadOrCropd(keys=["image", "label"], spatial_size=(256,256,128), mode=("constant")),
                Orientationd(keys=["image", "label"], axcodes="RAS"),
                ScaleIntensityRanged(
                    keys=["image"], a_min=-125, a_max=275,
                    b_min=0.0, b_max=1.0, clip=True,
                ),
                CropForegroundd(keys=["image", "label"], source_key='label', select_fn=lambda x: x > 0, margin=0),
                RandCropByPosNegLabeld(
                    keys=["image", "label"],
                    label_key="label",
                    spatial_size=(96, 96, 96),
                    pos=1,
                    neg=1,
                    num_samples=crop_samples,
                    allow_smaller=True,
                ),
                RandShiftIntensityd(
                    keys=["image"],
                    offsets=0.10,
                    prob=0.50,
                ),
                RandAffined(
                    keys=['image', 'label'],
                    mode=('bilinear', 'nearest'),
                    prob=1.0, spatial_size=(96, 96, 96),
                    rotate_range=(0, 0, np.pi / 30),
                    scale_range=(0.1, 0.1, 0.1)),
                ToTensord(keys=["image", "label"]),
            ]
        )

        val_transforms = Compose(
            [
                LoadImaged(keys=["image", "label"]),
                AddChanneld(keys=["image", "label"]),
                Spacingd(keys=["image", "label"], pixdim=(
                    1.0, 1.0, 1.2), mode=("bilinear", "nearest")),
                Orientationd(keys=["image", "label"], axcodes="RAS"),
                ScaleIntensityRanged(
                    keys=["image"], a_min=-125, a_max=275,
                    b_min=0.0, b_max=1.0, clip=True,
                ),
                CropForegroundd(keys=["image", "label"], source_key='label', select_fn=lambda x: x > 0, margin=0),
                ToTensord(keys=["image", "label"]),
            ]
        )

        test_transforms = Compose(
            [
                LoadImaged(keys=["image"]),
                AddChanneld(keys=["image"]),
                Spacingd(keys=["image"], pixdim=(
                    1.0, 1.0, 1.2), mode=("bilinear")),
                # ResizeWithPadOrCropd(keys=["image"], spatial_size=(168,168,128), mode=("constant")),
                Orientationd(keys=["image"], axcodes="RAS"),
                ScaleIntensityRanged(
                    keys=["image"], a_min=-125, a_max=275,
                    b_min=0.0, b_max=1.0, clip=True,
                ),
                CropForegroundd(keys=["image"], source_key="image"),
                ToTensord(keys=["image"]),
            ]
        )

The error that occurs is as follows:

Traceback (most recent call last):
  File "/home/3DUX-Net/main_train.py", line 257, in <module>
    global_step, dice_val_best, global_step_best = train(
  File "/home/3DUX-Net/main_train.py", line 217, in train
    dice_val = validation(epoch_iterator_val)
  File "/home/3DUX-Net/main_train.py", line 170, in validation
    dice_metric(y_pred=val_output_convert, y=val_labels_convert)
  File "/home/local/miniconda/envs/uxnet/lib/python3.9/site-packages/monai/metrics/metric.py", line 329, in __call__
    ret = super().__call__(y_pred=y_pred, y=y)
  File "/home/local/miniconda/envs/uxnet/lib/python3.9/site-packages/monai/metrics/metric.py", line 68, in __call__
    return self._compute_list(y_pred, y)
  File "/home/local/miniconda/envs/uxnet/lib/python3.9/site-packages/monai/metrics/metric.py", line 90, in _compute_list
    ret = [self._compute_tensor(p.detach().unsqueeze(0), y_.detach().unsqueeze(0)) for p, y_ in zip(y_pred, y)]
  File "/home/local/miniconda/envs/uxnet/lib/python3.9/site-packages/monai/metrics/metric.py", line 90, in <listcomp>
    ret = [self._compute_tensor(p.detach().unsqueeze(0), y_.detach().unsqueeze(0)) for p, y_ in zip(y_pred, y)]
  File "/home/local/miniconda/envs/uxnet/lib/python3.9/site-packages/monai/metrics/meandice.py", line 81, in _compute_tensor
    return compute_meandice(
  File "/home/local/miniconda/envs/uxnet/lib/python3.9/site-packages/monai/metrics/meandice.py", line 136, in compute_meandice
    raise ValueError(f"y_pred and y should have same shapes, got {y_pred.shape} and {y.shape}.")
ValueError: y_pred and y should have same shapes, got torch.Size([1, 2, 138, 62, 145]) and torch.Size([1, 2, 138, 62, 130]).

[KumoLiu]

Hi @MOMOANNIE, thanks for your interest here. I noticed that you cropped your training data to (96, 96, 96), you may also need to ensure your validation data to such size by using ResizeWithPadOrCropd.

Hope it helps, thanks!