Support Akoya data

[berombau]

Akoya has their new platform PhenoCycler Fusion, which output .qptiff files. Many datasets are publicly available. One example is the human tonsil dataset. This demo sample is probably also from the public human tonsil dataset. See the public dataset for an interactive demo.

• PhoneCycler-Fusion_demo.qptiff
• public dataset

aicsimageio has support to read the different channels in these image files.

When using napari, also install bioformats_jar in order to view the metadata correctly. There are 5 levels of resolutation: 1 high resolution to 5 low resolution. Other images include a thumbnail, overview image and label.

With the bioformats data, it is clear that the bottom layer in napari when viewing is the one for DAPI.

Remaining tasks:

• [x] read to xarray/ImageContainer
• [x] segment on DAPI
• [ ] somehow go from all these channel layers to cell identities. We possibly need more information (protein marker list) and can ask Akoya sales people.

[SilverViking]

It is also possible to read the .qptiff file with tifffile. This is a Python-only library which is being actively developed and does not rely on bioformats.jar. Via tifffile we can read the pyramidal image data, as well as metadata such as channel names and image resolution (pixel size).

I have example code in https://github.com/vibbits/spatial_pipeline/blob/main/experiments/qptiff-format.ipynb (private repo for now).

[lopollar]

Akoya support is out of scope for sparrow