`.tif` file from Ventana scanner is not parsed properly with WSIReader; openslide works well on its own

[GeorgeBatch]

• TIA Toolbox version: 1.5.1
• Python version: 3.11.8
• Operating System: CentOS Linux 7 (Core)

Description

I have slides scanned with a Ventana DP200 scanner and saved as .tif files.

What I Did

slide_path = "sample_slide_path.tif"

Open with tiatoolbox WSIReader and print the extracted metadata:

from tiatoolbox.wsicore.wsireader import WSIReader

reader = WSIReader.open(input_img=slide_path)
print(reader)

info_dict = reader.info.as_dict()
pprint(info_dict)  # noqa: T203

Result:

<tiatoolbox.wsicore.wsireader.VirtualWSIReader object at 0xABCDEFGHIJK>
{'axes': 'YSX',
 'file_path': PosixPath('sample_slide_path.tif'),
 'level_count': 1,
 'level_dimensions': ((1251, 3685),),
 'level_downsamples': [1.0],
 'mpp': (None, None),
 'objective_power': None,
 'slide_dimensions': (1251, 3685),
 'vendor': 'None'}

The result did not seem right, so I checked with openslide

slide_openslide = openslide.open_slide(slide_path)

print("openslide.level_downsamples", slide_openslide.level_downsamples)
print("openslide.level_dimensions", slide_openslide.level_dimensions)
print("openslide.mpp-x", slide_openslide.properties[openslide.PROPERTY_NAME_MPP_X])
print("openslide.mpp-y", slide_openslide.properties[openslide.PROPERTY_NAME_MPP_Y])
print("openslide.objective-power", slide_openslide.properties[openslide.PROPERTY_NAME_OBJECTIVE_POWER])

Result:

openslide.level_downsamples (1.0, 2.0, 4.0, 8.0, 16.0, 32.0, 64.0, 128.0, 256.0, 512.0, 1024.0)
openslide.level_dimensions ((90112, 134144), (45056, 67072), (22528, 33536), (11264, 16768), (5632, 8384), (2816, 4192), (1408, 2096), (704, 1048), (352, 524), (176, 262), (88, 131))
openslide.mpp-x 0.25
openslide.mpp-y 0.25
openslide.objective-power 40

For some reason, instead of using OpenSlideWSIReader or TIFFWSIReader, VirtualWSIReader is used. I will investigate further and add my findings in the comments below.

[GeorgeBatch]

I checked, last_suffix is ".tif", but is_tiled_tiff(slide_path) is False, so neither OpenSlideWSIReader nor TIFFWSIReader gets to be tried.

https://github.com/TissueImageAnalytics/tiatoolbox/blob/490be26bd422daf1e43ed98e974b56ac9bd878c3/tiatoolbox/wsicore/wsireader.py#L300-L304

I checked what happens in is_tiled_tiff when passing my slide_path. In the try-except block, a tif is successfully made with tif = tifffile.TiffFile(path), but tif.pages[0].is_tiled is False for my slide, so is_tiled_tiff(slide_path) is False.

https://github.com/TissueImageAnalytics/tiatoolbox/blob/490be26bd422daf1e43ed98e974b56ac9bd878c3/tiatoolbox/wsicore/wsireader.py#L62-L79

Given that openslide opens the image and successfully parses metadata, I think the requirement of is_tiled_tiff() to be true is too strict.

---

Finally, I checked out the develop branch with the tiatoolbox-dev environment created and activated as described in the README. The results are the same, which is as expected since the relevant parts of the code did not change:

https://github.com/TissueImageAnalytics/tiatoolbox/blob/b68c9ce21104a9153c078a6757d7f4b68dfa495a/tiatoolbox/wsicore/wsireader.py#L303-L307

and

https://github.com/TissueImageAnalytics/tiatoolbox/blob/b68c9ce21104a9153c078a6757d7f4b68dfa495a/tiatoolbox/wsicore/wsireader.py#L65-L82

[GeorgeBatch]

I made a workaround that I will make into a pull-request. Please let me know what you think of it:

Add an argument ignore_is_tiled_tiff: bool = False to wsicore.wsireader.WSIReader.open(): https://github.com/TissueImageAnalytics/tiatoolbox/blob/b68c9ce21104a9153c078a6757d7f4b68dfa495a/tiatoolbox/wsicore/wsireader.py#L231-L236

Change https://github.com/TissueImageAnalytics/tiatoolbox/blob/b68c9ce21104a9153c078a6757d7f4b68dfa495a/tiatoolbox/wsicore/wsireader.py#L303 to be

if last_suffix in (".tif", ".tiff") and (ignore_is_tiled_tiff or is_tiled_tiff(input_path)):

[GeorgeBatch]

@Abdol @shaneahmed, do you have an estimate of when you will release v1.5.2?

[shaneahmed]

Hi @GeorgeBatch , we plan to make a release v 1.6.0 as soon as #825 or #716 are ready. Probably by end of July/early Aug.

[GeorgeBatch]

@shaneahmed, thank you for the update!