How to generate dataset images?

[pccoe-co]

I am trying to use similar cell images dataset for 4 - Segment nuclei and cells.ipynb notebook. Can you tell me how to generate tiff images compatible with the notebook.

[muellerflorian]

hello, do you mean you would like to run the notebook on your own data?

@Henley13 , maybe we should add a section in the docs (if not present already) to describe how a user has to prepare his own data in the simplest possible way to use the notebooks?

[Henley13]

``Hello @pccoe-co,

I am not sure what you mean by generate tiff images, but all you need to run the notebook is to read your images in order to get a nucleus and cell image (preferably from DAPI and CellMask channels, respectively), in 2D, aligned and with the same shape. It does not matter if your image is saved in tiff, tif, png or anything else. You can use bigfish.stack.read_image to read your images for example.

More specifically, in the second code block you can use your own input and output paths and comment the checking line:

# hard-code the paths of our input and output directories
path_input = "../data/input"
path_output = "../data/output"

# check input images are loaded
#stack.check_input_data(path_input, input_segmentation=True)

In the third code block, you can rewrite the path with the name of you images:

path = os.path.join(path_input, "example_nuc_full.tif")
nuc = stack.read_image(path)
print("nucleus image")
print("\r shape: {0}".format(nuc.shape))
print("\r dtype: {0}".format(nuc.dtype), "\n")

path = os.path.join(path_input, "example_cell_full.tif")
cell = stack.read_image(path)
print("cell image")
print("\r shape: {0}".format(cell.shape))
print("\r dtype: {0}".format(cell.dtype))

[pccoe-co]

@muellerflorian yes I mean to say that I want to run this notebook on my own data. https://drive.google.com/drive/folders/14rsnAPYWUGVlMB5jiwF8rn-9PfWcOyq6?usp=sharing I am trying following image for segmentation.

[Henley13]

@pccoe-co you can read your jpg image with bigfish.stack.read_image. Have you tried to edit and run the code as explained here ?

Some parts of the notebook require cell and nucleus channels. Si if you only have a cell image you won't be able to fully run the notebook.