[2824d6]: / drunet / module.py

Download this file

171 lines (142 with data), 6.3 kB

  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
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
from collections import Counter
import pathlib
import math
import os
import matplotlib.pyplot as plt
import tensorflow as tf
import numpy as np
import cv2 as cv
import utils
import tqdm
def binary_image_from_dri(input_dir, threshold=128, save_dir=None):
if os.path.isdir(input_dir):
paths = utils.list_file(input_dir)
utils.check_file([save_dir])
else:
paths = [input_dir]
for path in paths:
path_stem = pathlib.Path(path).stem
image = cv.imread(path, 0)
bin_image = binary_image(image, threshold)
if save_dir is not None:
cv.imwrite(os.path.join(save_dir, '{}.jpg'.format(path_stem)), bin_image)
return
def binary_image(image, threshold):
shape = image.shape
if len(shape) == 3:
image = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
th, bin_image = cv.threshold(image, threshold, 255, cv.THRESH_BINARY)
return bin_image
def reverse_pred_image(normalize_pred_image):
reverse_image = normalize_pred_image.squeeze() * 255.0
reverse_image = np.array(reverse_image, dtype=np.uint8)
return reverse_image
def save_images(pred, index, save_path, image_shape, split=False):
image_numbers = int(np.sqrt(pred.shape[0]))
if not split:
h = image_shape[0]
w = image_shape[1]
H = int(image_numbers * image_shape[0])
W = int(image_numbers * image_shape[1])
big_image = np.zeros(shape=(H, W, image_shape[-1]), dtype=np.uint8).squeeze()
for i in range(pow(image_numbers, 2)):
image = (pred[i, :, :] * 255.0)
image = np.array(image, dtype=np.uint8)
image = image.squeeze()
j = i % image_numbers
k = i // image_numbers
if image_shape[-1] == 1 and len(image_shape) == 3:
big_image[k * h:(k + 1) * h, j * w:(j + 1) * w] = image
else:
big_image[k * h:(k + 1) * h, j * w:(j + 1) * w, :] = image
path = os.path.join(save_path, 'Segment_train_pred_{}.png'.format(index))
plt.imsave(path, big_image, cmap='gray')
else:
for i in range(image_numbers ** 2):
image = (pred[i, :, :] * 255.0)
image = np.array(image, dtype=np.uint8)
image = image.squeeze()
path = os.path.join(save_path, 'Segment_pred_{}_{}.png'.format(index, i))
plt.imsave(path, image, cmap='gray')
return
def get_area(image):
"""Count the area of bleeding area"""
image = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
_, bin_image = cv.threshold(image, 0, 255, cv.THRESH_BINARY)
count_result = Counter(list(bin_image.reshape(-1, )))
area = count_result.get(255)
return area
def pixel_to_ml(pixel_area, dpi=96):
if pixel_area is None:
pixel_area = 0.0
return pixel_area / pow(dpi, 2) * pow(25.4, 2) / 100
def draw_contours(image, mask, max_count=8, dpi=96):
image = np.array(image)
mask = np.array(mask)
height, width = image.shape[:2]
copy_image = image.copy()
if len(mask.shape) == 3 and mask.shape[-1] != 1:
mask = cv.cvtColor(mask, cv.COLOR_BGR2GRAY)
else:
mask = mask
mask = cv.resize(mask, (height, width))
th, bin_mask = cv.threshold(mask, 0, 255, cv.THRESH_BINARY)
con_list = []
blood_area = []
contours, _ = cv.findContours(bin_mask, cv.RETR_TREE, cv.CHAIN_APPROX_SIMPLE)
for index, contour in enumerate(contours):
area = cv.contourArea(contour)
if 200 < area < height * width * 0.94:
con_list.append(index)
blood_area.append(pixel_to_ml(area, dpi))
if len(con_list) > max_count:
blood_area = [0]
else:
for index in con_list:
copy_image = cv.drawContours(copy_image, contours, index, (0, 0, 255), 5)
return copy_image, sum(blood_area)
def save_invalid_data(origin_images, drawed_images, pred_mask_images, image_names, save_dir, reshape=True):
"""
:param image_names: the image file names of the original images, in the form of a list
:param origin_images: original bleeding images
:param drawed_images: draw a contour map of the bleeding area on the original image according to the mask
:param pred_mask_images: predicted mask image
:param save_dir: save path of all images
:param reshape: restore all images to the original image size
"""
origin_save_dir = os.path.join(save_dir, 'origin')
drawed_save_dir = os.path.join(save_dir, 'drawed')
mask_save_dir = os.path.join(save_dir, 'pred_mask')
utils.check_file([origin_save_dir, drawed_save_dir, mask_save_dir])
for index in range(len(origin_images)):
origin_image = origin_images[index]
drawed_image = drawed_images[index]
mask_image = pred_mask_images[index]
_, bin_mask_image = cv.threshold(mask_image, 0, 255, cv.THRESH_BINARY)
if reshape:
origin_image = cv.resize(origin_image, (256, 256))
drawed_image = cv.resize(drawed_image, (256, 256))
save_name = '{}'.format(image_names[index])
save_mask_path = os.path.join(mask_save_dir, save_name)
save_origin_path = os.path.join(origin_save_dir, save_name)
save_drawed_path = os.path.join(drawed_save_dir, save_name)
cv.imwrite(save_mask_path, bin_mask_image)
cv.imwrite(save_origin_path, origin_image)
cv.imwrite(save_drawed_path, drawed_image)
return
def count_volume(areas, thickness=0.45):
for area in areas:
if area == 0:
areas.remove(area)
areas_count = len(areas)
volume = [areas[index] * thickness for index in range(areas_count)]
return sum(volume)
def calculate_volume(mask_dir, real_shape=(1440, 1440), thickness=0.4, dpi=96):
all_areas = []
for path in tqdm.tqdm(pathlib.Path(mask_dir).iterdir()):
mask_image = cv.imread(str(path))
mask_image = cv.resize(mask_image, real_shape)
area = pixel_to_ml(get_area(mask_image), dpi=dpi)
all_areas.append(area)
volume = count_volume(all_areas, thickness)
return volume