-
Notifications
You must be signed in to change notification settings - Fork 0
/
create_tissue_masks.py
57 lines (48 loc) · 2.12 KB
/
create_tissue_masks.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
from __future__ import division, print_function
import os
import numpy as np
from PIL import Image as pil_image
import matplotlib.pyplot as plt
import cv2
import glob
def generate_background_masks(image_dir, mask_dir, my_palette):
kernel = np.ones((20, 20), 'uint8')
for fullname in glob.glob(image_dir + '/*.jpg'):
fname = fullname.split('\\')[-1].split('.')[0]
# read image as greyscale
tma_img = cv2.imread(fullname, 0)
# Gaussian filtering to remove noise
blur = cv2.GaussianBlur(tma_img, (25, 25), 0)
# Otsu thresholding (background should be assigned 0, tissue with 1)
ret, img_thres = cv2.threshold(blur, 0, 255, cv2.THRESH_BINARY_INV+cv2.THRESH_OTSU)
# add padding to avoid weird borders afterwards
bb = 100
img_thres = cv2.copyMakeBorder(img_thres,bb,bb,bb,bb,cv2.BORDER_CONSTANT,value=0)
# dilation to fill black holes
img = cv2.dilate(img_thres, kernel, iterations=5)
# followed by erosion to restore borders, eat up small objects
img = cv2.erode(img, kernel, iterations=10)
# then dilate again
img = cv2.dilate(img, kernel, iterations=5)
# crop to restore original image
ws = np.array(img)[bb:-bb, bb:-bb]
mask = np.zeros((tma_img.shape[0], tma_img.shape[1]), dtype='uint8')
mask[ws == 0] = 4
mask[ws == 255] = 0
hm = pil_image.fromarray(mask.astype('uint8'), 'P')
hm.putpalette(my_palette)
mask_name = os.path.join(mask_dir, 'mask_' + fname + '.png')
hm.save(mask_name)
def main():
my_palette = [0, 255, 0, # benign is green
0, 0, 255, # Gleason 3 is blue
255, 255, 0, # Gleason 4 is yellow
255, 0, 0, # Gleason 5 is red
255, 255, 255] # ignore class is white
pref = 'dataset_TMA'
image_dir = os.path.join(pref, 'TMA_images')
mask_dir = os.path.join(pref, 'tissue_masks')
os.makedirs(mask_dir)
generate_background_masks(image_dir, mask_dir, my_palette)
if __name__ == '__main__':
main()