Issue extracting cell information

[Boehmin]

First, such a great tool - I now managed to run cellpose and get cell and nuclei masks. I work with skeletal muscle tissue so I would also potentially have multiple nuclei per cell. I`m not sure if that might be the problem of the below issue:

1. I run the following code in a jupyter notebook:

fov_results = multistack.extract_cell(
    cell_label=cell_label, 
    ndim=2, 
    nuc_label=nuc_label, 
    rna_coord=spots_no_ts, 
    others_coord={"foci": foci, "transcription_site": ts},
    image=image_contrasted,
    others_image={"dapi": nuclei, "smfish": fish},
    check_nuc_in_cell=False) # IMPORTANT! otherwise it does not identify my cells
print("number of cells identified: {0}".format(len(fov_results)))

for i, cell_results in enumerate(fov_results):
    print("cell {0}".format(i))
    
    # get cell results
    cell_mask = cell_results["cell_mask"]
    cell_coord = cell_results["cell_coord"]
    nuc_mask = cell_results["nuc_mask"]
    nuc_coord = cell_results["nuc_coord"]
    rna_coord = cell_results["rna_coord"]
    foci_coord = cell_results["foci"]
    ts_coord = cell_results["transcription_site"]
    image_contrasted = cell_results["image"]
    print("\r number of rna {0}".format(len(rna_coord)))
    print("\r number of foci {0}".format(len(foci_coord)))
    print("\r number of transcription sites {0}".format(len(ts_coord)))
    
    # plot cell
    plot.plot_cell(
        ndim=2, cell_coord=cell_coord, nuc_coord=nuc_coord, 
        rna_coord=rna_coord, foci_coord=foci_coord, other_coord=ts_coord, 
        image=image_contrasted, cell_mask=cell_mask, nuc_mask=nuc_mask, 
        title="Cell {0}".format(i))

• sometimes it finds 29 cells (when I rerun the cell)
• it then sometimes finds 30 cells

2. As I keep rerunning extract cells and the plots, I sometimes get empty masks, sometimes the first cell has half of the image in it, and sometimes the background crops just fine.
   
   
   
   
3. However, the nuclei are not plotted and as is quite visible, the scale of the cells is off. I have no idea what this could be due to.
4. I had to also swap the coordinates for the saved spots and clusters - they did not match with the mask arrays, once I did this the cell extraction worked - I now wonder whether this messed up the scale later on, below my thought process:

print("Dimensions of mask array:", nuc_label.shape)
print("Dimensions of mask array:", cell_label.shape)

# Check maximum coordinate values
max_coords = np.max(spots, axis=0)
print("Maximum coordinate values:", max_coords)

# Check minimum coordinate values
min_coords = np.min(spots, axis=0)
print("Minimum coordinate values:", min_coords)

Dimensions of mask array: (1790, 1800)
Dimensions of mask array: (1790, 1800)
Maximum coordinate values: [1799 1789   17]
Minimum coordinate values: [ 0  0 -1]

# Swap x and y coordinates for spots
sw_spots = np.copy(spots)
sw_spots[:, 0], sw_spots[:, 1] = spots[:, 1], spots[:, 0]

# Swap x and y coordinates for clusters
sw_clusters = np.copy(clusters)
sw_clusters[:, 0], sw_clusters[:, 1] = clusters[:, 1], clusters[:, 0]

# Check maximum coordinate values
max_coords = np.max(sw_spots, axis=0)
print("Maximum coordinate values:", max_coords)

# Check minimum coordinate values
min_coords = np.min(sw_spots, axis=0)
print("Minimum coordinate values:", min_coords)

Maximum coordinate values: [1789 1799   17]
Minimum coordinate values: [ 0  0 -1]

5. This is how I saved the spots and clusters, I`m not sure whether there is a mistake in how they are saved. When I plot the spots, clusters and masks on top of each other I also have to flip the coordinates for the data to represent the original images. I load the .csv exactly as in the (6.) example notebook so I did not think I made a mistake there.

# save in csv files
path = os.path.join("test_spots.csv")
stack.save_data_to_csv(spots_post_clustering, path)
path = os.path.join("test_clusters.csv")
stack.save_data_to_csv(clusters, path)

Happy for any suggestions!