Confused on how to use this

[ridsmd]

Hello LIVECell creators, I'm trying to use your LIVECell code for a set of images of cells. By looking through your code, it seems you take images and compress them into Excel data, then a model runs through the Excel file data. I'm confused how my images should be converted to Excel data. Is that also done through your code, or is there a process I need to complete to convert my images? Also, I'm assuming the pre-processing code should be ran before the main code, but should the pre-processing code be ran in the same script as the main script, or should they be ran in separate scripts? If I could get a run-down of the process of using .tif images to the final product of using the model to predict what types of cells are present, that would be greatly appreciated. Thank you for any advice.

[RickardSjogren]

Hi @ridsmd, it is not clear to me what you are trying to achieve. Can you please explain what your inputs are and what outputs you are aiming to reach using the LIVECell code?

[ridsmd]

Hi Dr. Sjogren, thank you for your reply. I’m in a research lab at the UPMC Cancer Center. We have a large dataset of brightfield/phase microscopy images, they are .tif files. I read your article titled: LIVECell—A large-scale dataset for label-free live cell segmentation. Through this GitHub, I found some of your code. I wanted to use your code to go through our images and see if the model could segment and potentially annotate our images. Do you think it’s possible to run our dataset of images through the model you made in this article? And if so, how would I go about doing that? I see your “preprocessing” script and the main script, but I’m unsure of how I would use them on images. Is there some sort of conversion I’d have to do, where I turn the images into Excel data? How can I input the images into this model? For example, how were you able to run images through the script? I apologize if my explanation doesn’t make sense, I’m somewhat new to machine learning and python coding so if there’s parts you need more details for, please let me know. Thanks.

[rkjbs]

I am also quite confused on how to use this. I downloaded LIVECell-main and went through the very tedious process to get the whole LIVECell_dataset_2021 from AWS. I read the paper and all the documentation and README.MD files, and I unfortunately do not understand how to use either model.

I would like to, in the easiest way possible, to give the model my images of my cells in brightfield, and have it give me back results like those demonstrated in the paper (primarily interested in cell counts and areas).

Another point: I don't understand the difference between the anchor-free and anchor-based models, and when to use which. But that's a moot point at the moment as I can't figure out how to use either

[rkjbs]

I have now successfully installed all of the requirements: Python >3.6, PyTorch, torchvision, pycocotools, and detectron-ResNeSt. I do not know how to proceed from there to actually using the models based on the .yaml files in LIVECell-main or the .pth files in LIVECell_dataset_2021 nor how to run any of the .py files (including preprocessing.py)