Cannot find image metadata on 'Generative Modeling with Conditional Autoencoders: Building an Integrated Cell' dataset

Hi, thank you for making available all of the beautiful datasets!

I am trying to pre-process the dataset found in ‘Generative Modeling with Conditional Autoencoders:
Building an Integrated Cell’ pre-print. I believe it is the dataset found at the very bottom of this page: Download cell data (images, genomics, features) - ALLEN CELL EXPLORER

However, I cannot seem to be able to find the metadata regarding pixel size for each ‘.ome.tif’. This is important’.ome.tif’ files do not seem to have the same image dimensions (most are [1024, 1024], others are [624, 924]; [H, W]) and I would like to ensure that all resulting single-cell images are of the same micron dimensions.

Any help and information would be greatly appreciated! Once again thank you for the hard work!


The paper you named references this github repository: GitHub - AllenCellModeling/torch_integrated_cell which is using a 2D dataset. The readme in that repository gives instructions on how to download the training data.

It’s a little bit hard to tell what data files you have downloaded - are you trying to prepare the 2d data from the original 3d images? Then you will need to scale using the ome metadata to normalize for pixel sizes.
For the 3d tiff images you are referring to, if you could give an example filename for the 1024x1024 image and an example of the 624x924 image it would help us tell why they are different.