Image Analysis, Metadata, and Data Management

Hello everyone,

I have a few questions regarding image analysis and the best practice convention.

I have sets of neuronal images in the ome tiff format that has three channels per image with metadata.
What I have been doing is that I split the channels to use them in CellProfiler to correct illumination.
However, when I split the channels in python and save them again in .tif format, I do not tag the metadata along with the images.
Now, my questions are:

  1. Does CellProfiler or Noise2Void use information contained in the metadata of images? Because I do not tag the original metadata along with the split images, the one gets generated upon saving in .tif format contains less information than the original metadata.

  2. Do you guys think that it is important to keep the original metadata as “metadata” with the images? Or if only the information in the metadata is important, then should I just keep that information in the description of tif images?

  3. So, I also tried to manipulate the original metadata to reflect the channeled-images. For example, when I split an image, I also delete other channel information from the original metadata. Then I tried to save the single-channel image with the new metadata that only has information about that single channel, but aicsimage.ometiffwriter does not allow me to save an image in ome tiff format that has less than 3 dimensions.
    Is there a way to save a single-channel image in ome-tiff format? Or does it have to be 3 or more channels?

  4. How do you guys generally handle images for analysis? For multi-channel, should I keep it as is when using CellProfiler? I guess I could load a multi-channel images, but split them in CP, do the illumination correction, then put them back in one image to save?

Any advice would be greatly appreciated!!!