Reading Pathology Slides using the LoadImage functionality #3853

vikashg · 2022-02-24T07:54:39Z

vikashg
Feb 24, 2022
Collaborator

Hello everyone,
I am trying to read Whole Slide Images following the example code in MIL example.

LoadImageD(keys=["image"], reader=WSIReader, backend="TiffFile", dtype=np.uint8, level=1, image_only=True),

However, my slides are in svs format and I am using the cuCIM backend

train_transform = LoadImage(reader=WSIReader, backend='cucim')

I get the following error

[Plugin: cucim.kit.cuslide] Loading the dynamic library from: /opt/conda/lib/python3.8/site-packages/cucim/clara/[email protected]
Initializing plugin: cucim.kit.cuslide (interfaces: [cucim::io::IImageFormat v0.1]) (impl: cucim.kit.cuslide)
[Plugin: cucim.kit.cumed] Loading the dynamic library from: /opt/conda/lib/python3.8/site-packages/cucim/clara/[email protected]
Initializing plugin: cucim.kit.cumed (interfaces: [cucim::io::IImageFormat v0.1]) (impl: cucim.kit.cumed)
TIFFFetchDirectory: /workspace/data/train_and_validation_data/Train/WSIs/Cancerous/DCIS/BRACS_1999.svs: Can not read TIFF directory count.
TIFFReadDirectory: Failed to read directory at offset 1984839918.
Traceback (most recent call last):
  File "ExtractPatches.py", line 24, in <module>
    main()
  File "ExtractPatches.py", line 19, in main
    image = train_transform([_filepath])
  File "/workspace/MONAI/monai/transforms/io/array.py", line 226, in __call__
    raise RuntimeError(
RuntimeError: LoadImage cannot find a suitable reader for file: /workspace/data/train_and_validation_data/Train/WSIs/Cancerous/DCIS/BRACS_1999.svs.
    Please install the reader libraries, see also the installation instructions:
    https://docs.monai.io/en/latest/installation.html#installing-the-recommended-dependencies.
   The current registered: [<monai.data.image_reader.NumpyReader object at 0x7fa820e2a280>, <monai.data.image_reader.PILReader object at 0x7fa820d9b730>, <monai.data.image_reader.WSIReader object at 0x7fa820d9bca0>].
Traceback (most recent call last):
  File "/workspace/MONAI/monai/transforms/io/array.py", line 211, in __call__
    img = reader.read(filename)
  File "/workspace/MONAI/monai/data/image_reader.py", line 785, in read
    img = self.wsi_reader(name, **kwargs_)
ValueError: Cannot load /workspace/data/train_and_validation_data/Train/WSIs/Cancerous/DCIS/BRACS_1999.svs!

Traceback (most recent call last):
  File "/workspace/MONAI/monai/transforms/io/array.py", line 211, in __call__
    img = reader.read(filename)
  File "/workspace/MONAI/monai/data/image_reader.py", line 662, in read
    img = PILImage.open(name, **kwargs_)
TypeError: open() got an unexpected keyword argument 'backend'

Traceback (most recent call last):
  File "/workspace/MONAI/monai/transforms/io/array.py", line 211, in __call__
    img = reader.read(filename)
  File "/workspace/MONAI/monai/data/image_reader.py", line 572, in read
    img = np.load(name, allow_pickle=True, **kwargs_)
TypeError: load() got an unexpected keyword argument 'backend'

[Plugin: cucim.kit.cuslide] Unloaded.
[Plugin: cucim.kit.cumed] Unloaded.

I wonder why it is searching the TIFFFetchDirectory,

On the other hand, if the use the WSIReader from monai.data. I can read the files properly. So the following code works

import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)

# Install monai using the instructions at https://monai.io/started.html#installation or using the pip command `pip install monai`
import monai
from monai.data import WSIReader
from monai.transforms import LoadImage
from pathlib import Path
import numpy as np
from PIL import Image
filename='/workspace/data/Cancerous/DCIS/BRACS_1238.svs' #Any WSI 
reader = WSIReader(backend='cuCIM')
img_obj = reader.read(data=filename)

I wonder what I am doing wrong in the LoadImage functionality. Can someone help. Thanks

Answered by bhashemian

Feb 28, 2022

Hi @vikashg,
I cannot reproduce the issue using the image that you provided me with. Could you please check again using the following code?

from monai.transforms import LoadImaged, LoadImage
from monai.data.image_reader import WSIReader                                                                                                                                                                                                
                                                                                                                                                                                                                                                 
image_path = "./BRACS_1244…

View full answer

bhashemian · 2022-02-24T20:24:52Z

bhashemian
Feb 24, 2022
Collaborator

Hi @vikashg, thanks for sharing your question. In the approach where you are calling WSIReader directly, can you check if you can get the image data too?

img_array, meta_data = reader.get_data(img_obj)

3 replies

vikashg Feb 25, 2022
Collaborator Author

@drbeh Yes I can get the image data as a numpy array. Actually my goal was to extract patches at different zoom levels. So I was looking at MIL example

    valid_transform = Compose(
        [
            LoadImageD(keys=["image"], reader=WSIReader, backend="TiffFile", dtype=np.uint8, level=1, image_only=True),
            LabelEncodeIntegerGraded(keys=["label"], num_classes=args.num_classes),
            TileOnGridd(
                keys=["image"],
                tile_count=None,
                tile_size=args.tile_size,
                random_offset=False,
                background_val=255,
                return_list_of_dicts=True,
            ),
            ScaleIntensityRangeD(keys=["image"], a_min=np.float32(255), a_max=np.float32(0)),
            ToTensord(keys=["image", "label"]),
        ]
    )

I am interested in using the TileOnGrid. If I use the WSIReader directly. I cant make the pipeline as shown.

bhashemian Feb 25, 2022
Collaborator

Can you share your svs file so we can check on this?

vikashg Feb 25, 2022
Collaborator Author

You can download one example file here Its taken from the BRIGHT challenge hosted by ISBI

bhashemian · 2022-02-28T15:19:45Z

bhashemian
Feb 28, 2022
Collaborator

Hi @vikashg,
I cannot reproduce the issue using the image that you provided me with. Could you please check again using the following code?

from monai.transforms import LoadImaged, LoadImage
from monai.data.image_reader import WSIReader                                                                                                                                                                                                
                                                                                                                                                                                                                                                 
image_path = "./BRACS_1244.svs"

transform = LoadImage(reader=WSIReader, backend='cucim', level=3)
image, _ = transform(image_path)
print(f"LoadImage : image shape = {image.shape}")

transformD = LoadImaged(keys=["image"], reader=WSIReader, backend='cucim', level=3)
image = transformD({"image": image_path})["image"]
print(f"LoadImageD: image shape = {image.shape}")

3 replies

vikashg Feb 28, 2022
Collaborator Author

So, it did work but again when I tried on another image from the same dataset It didn't. For example this image

I get the following error

root@30ec0f062b57:/workspace/app# python ExtractPatches.py 
/workspace/data/train_and_validation_data/Train/WSIs/Cancerous/DCIS/BRACS_1999.svs
[Plugin: cucim.kit.cuslide] Loading the dynamic library from: /opt/conda/lib/python3.8/site-packages/cucim/clara/[email protected]
Initializing plugin: cucim.kit.cuslide (interfaces: [cucim::io::IImageFormat v0.1]) (impl: cucim.kit.cuslide)
[Plugin: cucim.kit.cumed] Loading the dynamic library from: /opt/conda/lib/python3.8/site-packages/cucim/clara/[email protected]
Initializing plugin: cucim.kit.cumed (interfaces: [cucim::io::IImageFormat v0.1]) (impl: cucim.kit.cumed)
TIFFFetchDirectory: /workspace/data/train_and_validation_data/Train/WSIs/Cancerous/DCIS/BRACS_3302.svs: Can not read TIFF directory count.
TIFFReadDirectory: Failed to read directory at offset 3165050832.
Traceback (most recent call last):
  File "ExtractPatches.py", line 23, in <module>
    main()
  File "ExtractPatches.py", line 19, in main
    X = train_transform(filename)		
  File "/workspace/MONAI/monai/transforms/io/array.py", line 209, in __call__
    raise RuntimeError(
RuntimeError: cannot find a suitable reader for file: /workspace/data/train_and_validation_data/Train/WSIs/Cancerous/DCIS/BRACS_3302.svs.
    Please install the reader libraries, see also the installation instructions:
    https://docs.monai.io/en/latest/installation.html#installing-the-recommended-dependencies.
   The current registered: [<monai.data.image_reader.ITKReader object at 0x7f7d256a34f0>, <monai.data.image_reader.NumpyReader object at 0x7f7d256f8730>, <monai.data.image_reader.PILReader object at 0x7f7d256f8d30>, <monai.data.image_reader.NibabelReader object at 0x7f7d256f8c40>, <monai.data.image_reader.WSIReader object at 0x7f7d2569d910>].

[Plugin: cucim.kit.cuslide] Unloaded.
[Plugin: cucim.kit.cumed] Unloaded.
root@30ec0f062b57:/workspace/app#

This is the code I used:

from monai.apps.pathology.transforms import TileOnGrid
from monai.data.image_reader import WSIReader
#from monai.data import WSIReader
from monai.transforms import Transform, Compose, LoadImage
import os
#from monai.apps.pathology.data.datasets import PatchWSIDataset



def main():
	data_dir = '/workspace/data'
	svs_data = os.path.join(data_dir, 'train_and_validation_data/Train/WSIs/Cancerous/DCIS')
	list_files = os.listdir(svs_data)
	all_list_files = [os.path.join(svs_data, _) for _ in list_files]
	data_dict={"image": all_list_files}
	train_transform = LoadImage( reader=WSIReader, backend='cucim', level=3)
	print(all_list_files[0])
	filename = '/workspace/data/train_and_validation_data/Train/WSIs/Cancerous/DCIS/BRACS_3302.svs'
	X = train_transform(filename)		


if __name__ == '__main__':
	main()

bhashemian Mar 1, 2022
Collaborator

@vikashg, that should be a problem with the file. None of the backends can load it (even when applied directly) and they all say that there is an invalid offset in the tiff file.

<tifffile.TiffPages @1984839918> invalid offset to first page 1984839918

vikashg Mar 1, 2022
Collaborator Author

@drbeh Thanks for checking. I guess there was problems with the files in the competition itself.
Thanks once again.
Vikash

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Reading Pathology Slides using the LoadImage functionality #3853

Uh oh!

{{title}}

Uh oh!

Replies: 2 comments 6 replies

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{editor}}'s edit

{{editor}}'s edit

Uh oh!

Uh oh!

{{title}}

Uh oh!

Select a reply

Uh oh!

Reading Pathology Slides using the LoadImage functionality #3853

Uh oh!

vikashg Feb 24, 2022 Collaborator

Replies: 2 comments · 6 replies

Uh oh!

bhashemian Feb 24, 2022 Collaborator

Uh oh!

vikashg Feb 25, 2022 Collaborator Author

Uh oh!

bhashemian Feb 25, 2022 Collaborator

Uh oh!

vikashg Feb 25, 2022 Collaborator Author

Uh oh!

bhashemian Feb 28, 2022 Collaborator

Uh oh!

vikashg Feb 28, 2022 Collaborator Author

Uh oh!

Uh oh!

bhashemian Mar 1, 2022 Collaborator

Uh oh!

vikashg Mar 1, 2022 Collaborator Author

vikashg
Feb 24, 2022
Collaborator

Replies: 2 comments 6 replies

bhashemian
Feb 24, 2022
Collaborator

vikashg Feb 25, 2022
Collaborator Author

bhashemian Feb 25, 2022
Collaborator

vikashg Feb 25, 2022
Collaborator Author

bhashemian
Feb 28, 2022
Collaborator

vikashg Feb 28, 2022
Collaborator Author

bhashemian Mar 1, 2022
Collaborator

vikashg Mar 1, 2022
Collaborator Author