Can a 3D numpy array be converted into a 3D image in Python?

Question

I made the following virtual "room" in a 3D array and would like to visualize it. I can't find a way to do so, please assist. The idea is to see a "3D image" of the array as a plot where the different values have different colours or just greyscale intensities, so that you can see the "patient" and the "detector" inside the "room":

import numpy as np


# Diff. values in the Room define objects: 0 = walls, 1 = inside room, 2 = patient's tissue, 3 = bone, 4 = x-ray detector

Room = np.array([[[0.0 for i in range(0,101,1)] for j in range(0,101,1)] for k in range(0,101,1)]) #The entire room with walls

for i in range(1,100,1):
    for j in range(1,100,1):
        for k in range(1,100,1):
            Room[i,j,k] +=1     # The room not counting the walls

for i in range(30,70,1):
    for j in range(30,70,1):
        for k in range(30,70,1):
            Room[i,j,k] +=1      #The patient's body (tissue)
            
for i in range(50,55,1):
    for j in range(50,55,1):
        for k in range(50,55,1):
            Room[i,j,k] +=1      #The patient's bone #1 
            
for i in range(58,63,1):
    for j in range(58,63,1):
        for k in range(58,63,1):
            Room[i,j,k] +=1      #The patient's bone #2

for i in range(88,92,1):
    for j in range(10,90,1):
        for k in range(10,90,1):
            Room[i,j,k] +=1      # X-ray Detector

Answer 1

You can create a 3 dimensional mesh grid with the help of matplotlib and numpy. Here is an example of such a plot. You just want to feed in your X,Y, and Z values as lists

import numpy as np
import matplotlib.pyplot as plt

# Create figure and add axis
fig = plt.figure(figsize=(8,6))
ax = plt.subplot(111, projection='3d')
    
# Remove gray panes and axis grid
ax.xaxis.pane.fill = False
ax.xaxis.pane.set_edgecolor('white')
ax.yaxis.pane.fill = False
ax.yaxis.pane.set_edgecolor('white')
ax.zaxis.pane.fill = False
ax.zaxis.pane.set_edgecolor('white')
ax.grid(False)
# Remove z-axis
ax.w_zaxis.line.set_lw(0.)
ax.set_zticks([])
    
# Create meshgrid
X, Y = np.meshgrid(np.linspace(0, 2, len(afm_data)), np.linspace(0, 2, len(afm_data)))
    
# Plot surface
plot = ax.plot_surface(X=X, Y=Y, Z=Z, cmap='YlGnBu_r', vmin=0, vmax=200)

There is also an article on this topic by Towards Data Science: https://towardsdatascience.com/visualizing-three-dimensional-data-heatmaps-contours-and-3d-plots-with-python-bd718d1b42b4

Answer 2

Just re-adding this because the first answer got deleted before I had the chance to respond:

Are you looking for a volume rendering approach? There's a broad range of Python libraries with that functionality available. This is an example link.

For your concrete case, you can use e.g. plotly and I just wrote a quick script for your data based on their tutorial:

import numpy as np
import plotly.graph_objects as go

# Generate nicely looking random 3D-field
np.random.seed(0)
l = 5
X, Y, Z = np.mgrid[:l, :l, :l]
vol = np.zeros((l, l, l))
#pts = (l * np.random.rand(3, 15)).astype(np.int)


# Diff. values in the Room define objects: 0 = walls, 1 = inside room, 2 = patient's tissue, 3 = bone, 4 = x-ray detector

Room = np.array([[[0.0 for i in range(0,102,1)] for j in range(0,102,1)] for k in range(0,102,1)]) #The entire room with walls

for i in range(1,100,1):
    for j in range(1,100,1):
        for k in range(1,100,1):
            Room[i,j,k] +=1     # The room not counting the walls

for i in range(30,70,1):
    for j in range(30,70,1):
        for k in range(30,70,1):
            Room[i,j,k] +=1      #The patient's body (tissue)
            
for i in range(50,55,1):
    for j in range(50,55,1):
        for k in range(50,55,1):
            Room[i,j,k] +=1      #The patient's bone #1 
            
for i in range(58,63,1):
    for j in range(58,63,1):
        for k in range(58,63,1):
            Room[i,j,k] +=1      #The patient's bone #2

for i in range(88,92,1):
    for j in range(10,90,1):
        for k in range(10,90,1):
            Room[i,j,k] +=1      # X-ray Detector

pts = Room.reshape(3,353736).astype(np.int)

print( tuple(indices for indices in pts) )

vol[tuple(indices for indices in pts)] = 1

from scipy import ndimage

vol = ndimage.gaussian_filter(vol, 0.5)
vol /= vol.max()

fig = go.Figure(data=go.Volume(
    x=X.flatten(), y=Y.flatten(), z=Z.flatten(),
    value=vol.flatten(),
    isomin=0.2,
    isomax=0.7,
    opacity=0.1,
    surface_count=25,
    ))
fig.update_layout(scene_xaxis_showticklabels=False,
                  scene_yaxis_showticklabels=False,
                  scene_zaxis_showticklabels=False)
fig.show()

The output is given below. Not sure if this is what you had in mind.