Extracting information about the 3D roi and their naming logic

I believe the easiest way is to expert the ROI map from the GUI. Without binarization, you will get an ZxYxX tiff file, that has 0 for background and then 1 for ROI 1, 2 ROI 2, etc. A 3D ROI can span multiple planes. Some example code with help of ChatGPT:

import numpy as np
from scipy.ndimage import label, center_of_mass

# Function to create a sphere mask
def create_sphere(radius):
    diameter = 2 * radius + 1
    cube = np.zeros((diameter, diameter, diameter), dtype=np.uint8)
    center = radius
    for x in range(diameter):
        for y in range(diameter):
            for z in range(diameter):
                if (x - center) ** 2 + (y - center) ** 2 + (z - center) ** 2 <= radius ** 2:
                    cube[x, y, z] = 1
    return cube

# Function to place a sphere in the cube at a specified position
def place_sphere(cube, sphere, position):
    x, y, z = position
    x_end = min(x + sphere.shape[0], cube.shape[0])
    y_end = min(y + sphere.shape[1], cube.shape[1])
    z_end = min(z + sphere.shape[2], cube.shape[2])
    cube[x:x_end, y:y_end, z:z_end] += sphere[:x_end-x, :y_end-y, :z_end-z]
    return cube

# Create a 150x150x150 cube
cube = np.zeros((150, 150, 150), dtype=np.uint8)

# Create spheres and place them in the cube
sphere_radii = [5, 8, 10, 12]
for i, radius in enumerate(sphere_radii):
    sphere = create_sphere(radius)
    position = (20*(i+1), 20*(i+1), 20*(i+1))  # Adjust position to avoid overlap
    cube = place_sphere(cube, sphere * (i+1), position)

# Get the volume and center of mass of each sphere
### THESE ARE THE RELEVANT LINES
for i, radius in enumerate(sphere_radii):
    sphere_id = i + 1
    sphere_mask = cube == sphere_id
    volume = np.sum(sphere_mask)
    com = center_of_mass(sphere_mask)
    print(f"Sphere {sphere_id}: Volume = {volume}, Center of Mass = {com}")

The last couple of lines are the relevant ones.

Here's the output:

Sphere 1: Volume = 515, Center of Mass = (25.0, 25.0, 25.0)
Sphere 2: Volume = 2109, Center of Mass = (48.0, 48.0, 48.0)
Sphere 3: Volume = 4169, Center of Mass = (70.0, 70.0, 70.0)
Sphere 4: Volume = 7153, Center of Mass = (92.0, 92.0, 92.0)

And the image of the spheres as z-projection (some "fake" dendritic spines")

ankilab / DeepD3

Extracting information about the 3D roi and their naming logic #6