preprocessing_scripts/preprocess_camelyon.py

import os
import slideio
import numpy as np
from PIL import Image
from tqdm import tqdm

data_directory = "/vol/data/histo_datasets/CAMELYON/CAMELYON17/"
mask_list = os.listdir(data_directory + "masks/")
for m in tqdm(mask_list):
    slide = slideio.open_slide(data_directory + "masks/" + m)
    image_slide = slideio.open_slide(data_directory + "images/" + m[:-9] + ".tif")
    scene = slide.get_scene(0)
    image_scene = image_slide.get_scene(0)

    dim0 = int(np.ceil(scene.size[0] / 1024))
    dim1 = int(np.ceil(scene.size[1] / 1024))
    resolutions = np.ceil(np.log2(max(dim0,dim1)))
    for r in range(int(resolutions) + 1):
        res = 2**r * 1024
        dim0 = int(np.ceil(scene.size[0] / res))
        dim1 = int(np.ceil(scene.size[1] / res))
        last_dim = (int(scene.size[0] % res), int(scene.size[1] % res))

        if last_dim[0] == 0 and last_dim[1] == 0:
            last_dim = (res, res)
        elif last_dim[0] == 0:
            last_dim = (res, last_dim[1])
        elif last_dim[1] == 0:  
            last_dim = (last_dim[0], res)

        for i in range(dim0):
            for j in range(dim1):
                if i == dim0-1 and j == dim1-1:
                    width = last_dim[0]
                    height = last_dim[1]
                elif i == dim0-1:
                    width = last_dim[0]
                    height = res
                elif j == dim1-1:
                    width = res
                    height = last_dim[1]
                else:
                    width = res
                    height = res
                mask = scene.read_block((i*res,j*res, width, height), (width // (2**r), height // (2**r)))
                mask = np.where(mask == 2, 1, 0).astype(np.uint8)
                if (np.max(mask) == 1):
                    image = image_scene.read_block((i*res,j*res, width, height), (width // (2**r), height // (2**r)))
                    # Save image and mask
                    # Save image
                    Image.fromarray(image).save(data_directory + "images_patches/" + m[:-9] + "_{}_{}_{}_{}_{}_{}.png".format(i*res,j*res,width, height, width // (2**r),  height // (2**r)))
                    # Save mask
                    Image.fromarray(mask).save(data_directory + "masks_patches/" + m[:-9] + "_{}_{}_{}_{}_{}_{}.png".format(i*res,j*res,width, height, width // (2**r),  height // (2**r)))