search

AllenCellModeling / pytorch_fnet

Three dimensional cross-modal image inference
Other
151 stars 61 forks source link

question about the data #177

Closed Cptmaxon closed 3 years ago

Cptmaxon commented 3 years ago

I'm trying to understand the metadata.csv file from quilt. but the readme doesn't really explain what the FOV image contains. it has 7 channels, I understand one is brightfield and the other is the structure , but what about the extra 5? thanks in advance if you can point me to someone who can help , that would be amazing.

evamaxfield commented 3 years ago

Hey @Cptmaxon

From the README: "Each field of view (FOV) includes 4 channels (BF, EGFP, DNA, Cell membrane)"

However, that does indeed miss that there are seven channels in some images. These are extra dummy channels added during image acquisition.

We recommend using the metadata for each image to know which channel is which, with aicsimageio that is as simple as:

from aicsimageio import AICSImage

img = AICSImage("my_file.ome.tiff")
img.get_channel_names()

Link to get_channel_names doc

(Use whatever image reading library you would like, I just know aicsimageio best as I am a maintainer)

You will be able to know which channels are dummy channels both visibly when loaded in an image viewer or if they have an _ in their channel name metadata value, i.e. H3342 is the valid channel for Hoechst dye to stain DNA, whileH3342_1 or H3342_2 -- the _1 or _2 mean it is a dummy channel.

Hope this helps! Let me know if you have any other questions.

Cptmaxon commented 3 years ago

hmm haven't seen that in the readme, thank you so much!

evamaxfield commented 3 years ago

Sorry. By README I meant the Quilt data package README not this repo. My bad!