Generating Magnetic Resonance images using DALL-E#

In this notebook we will demonstrate how to ask openAI’s DALL-E model to generate medial images (for fun).

Read more:

import openai
from skimage.io import imread, imshow
from numpy import random
from matplotlib import pyplot as plt

def prompt_image(message:str, width:int=1024, height:int=1024, model='dall-e-3'):
    client = openai.OpenAI()
    response = client.images.generate(
      prompt=message,
      model=model,
      n=1,
      size=f"{width}x{height}"
    )
    image_url = response.data[0].url
    image = imread(image_url)
    
    return image

As a real example of an orange slice imaged with MR, we use the example dataset “Credit: Mandarin orange, axial view, MRI.” is licensed (CC-BY 4.0) by Alexandr Khrapichev, University of Oxford

images = [imread('../../data/mri_fruit_sxm89b3x.jpg')[3063:4087,1024:2048,0]]

mri_prompt = """
A single, high resolution, black-white image of 
a realistically looking orange fruit slice 
imaged with T2-weighted magnetic resonance imaging (MRI).
"""

for _ in range(3):
    images.append(prompt_image(mri_prompt))

random.shuffle(images)

fix, ax = plt.subplots(1,len(images), figsize=(15,15))
for i, image in enumerate(images):
    ax[i].imshow(image, cmap='Greys_r')
../_images/fe2ca99ef91f5ba11806bd780256a40b392d9a2dd08cb4f26bb19539d60eabea.png

Exercise#

There is another example data set available. Crop out the star fruit from that image and repeat the experiment: Write a prompt that generates images looking similar.

The example dataset “Collage of mixed fruits and vegetables, MRI.” is licensed (CC-BY 4.0) by Alexandr Khrapichev, University of Oxford

image2 = imread('../../data/mri_fruit_bvtnk4mm.jpg')
imshow(image2)

<matplotlib.image.AxesImage at 0x10a35ff4ac0>
../_images/e8ba58e6d88b7a94e0ad3dd343402d4f12dd8d6642bb77c33cc14b0e02e218d3.png