-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathextraction.py
192 lines (155 loc) · 6.96 KB
/
extraction.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
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
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
import numpy as np
import cv2
from matplotlib import pyplot as plt
import pandas as pd # for reading and writing tables
import ntpath
from numpy import where
from collections import Counter
# Import utility functions
from util import *
# Define folder from which data can be downloaded
data_folder = "../training_set/"
benign_folder = data_folder + "benign/"
malignant_folder = data_folder + "malignant/"
# Get count of images
benign, malignant, n_ben, n_mal = getCountOfImage(data_folder)
# Output Log
print("Counted", benign, "benign,", malignant, "malignant: \n-", n_ben, "total benign\n-", n_mal, "total malignant")
size_x ,size_y = 128,128
#create empty array of zeros to store image inside it
study_benign = np.zeros((benign,size_x,size_y))
mask_benign = np.zeros((benign,size_x,size_y))
study_malignant = np.zeros((malignant,size_x,size_y))
mask_malignant = np.zeros((malignant,size_x,size_y))
labels_benign = np.zeros(benign)
labels_malignant = np.zeros(malignant)
correspondence_index_ben = np.zeros((n_ben,1))
correspondence_index_mal = np.zeros((n_mal,1))
#start load image
#0 #1
classes = ['benign', 'malignant']
label = 0
labels = [] #for classification part
images = [] #for classification part
i_ben = 0
i_mal = 0
for cname in os.listdir(data_folder):
for filename in sorted (os.listdir(os.path.join(data_folder,cname))):
imagePath = data_folder + cname + '/' + filename
image = cv2.imread(imagePath,cv2.IMREAD_GRAYSCALE)
if not '_mask' in filename :
image = cv2.resize(image,(size_x,size_y))
image = np.array(image)
images.append(image)
if 'benign' in filename:
correspondence_index_ben[num(filename)-1] = i_ben
study_benign[i_ben]+= np.array(image)
labels_benign[i_ben] = int(0)
i_ben +=1
if 'malignant' in filename:
correspondence_index_mal[num(filename)-1] = i_mal
study_malignant[i_mal]+= np.array(image)
labels_malignant[i_mal] = int(1)
i_mal +=1
for filename in sorted (os.listdir(os.path.join(data_folder,cname))):
imagePath = data_folder + cname + '/' + filename
image = cv2.imread(imagePath,cv2.IMREAD_GRAYSCALE)
if '_mask' in filename :
image = cv2.resize(image,(size_x,size_y))
image = np.array(image)
if filename[0] == 'b':
ind_ = int(correspondence_index_ben[num(filename)-1])
mask_benign[ind_]+= np.array(image)
if filename[0] == 'm' :
ind_ = int(correspondence_index_mal[num(filename)-1])
mask_malignant[ind_]+= np.array(image)
mask_benign[mask_benign > 0] = 1
mask_malignant[mask_malignant > 0] = 1
print("\nImported:")
print("- benign shape:", np.shape(study_benign), "\tmask:", np.shape(mask_benign))
print("- malignant shape:", np.shape(study_malignant), "\tmask:", np.shape(mask_malignant))
# Collect studies together
studies = (np.concatenate((study_benign, study_malignant), axis = 0))
masks = (np.concatenate((mask_benign, mask_malignant), axis = 0))
labels = (np.concatenate((labels_benign, labels_malignant), axis = 0))
# import useful packages
from radiomics import featureextractor
# special functions for using pyradiomics
from SimpleITK import GetImageFromArray
import radiomics
from radiomics.featureextractor import RadiomicsFeatureExtractor # This module is used for interaction with pyradiomic
import logging
logging.getLogger('radiomics').setLevel(logging.CRITICAL + 1) # this tool makes a whole TON of log noise
# Plot Setup Code
# Setup the defaults to make the plots look a bit nicer for the notebook
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
import seaborn as sns
plt.rcParams["figure.figsize"] = (8, 8)
plt.rcParams["figure.dpi"] = 125
plt.rcParams["font.size"] = 14
plt.rcParams['font.family'] = ['sans-serif']
plt.rcParams['font.sans-serif'] = ['DejaVu Sans']
plt.style.use('ggplot')
sns.set_style("whitegrid", {'axes.grid': False})
# Define extractor
extractor = featureextractor.RadiomicsFeatureExtractor(binCount = 128, force2D = True)
extractor.enableFeatureClassByName('shape2D')
extractor.settings
results = extractor.execute(GetImageFromArray(studies[0]), GetImageFromArray((masks[0]).astype(np.uint8)), label = 1)
# Extract features from Benign and Malignant studies
for i in range(len(studies)):
features_currentStudy = extractor.execute(GetImageFromArray(studies[i]), GetImageFromArray((masks[i]).astype(np.uint8)), label = 1)
# Stack DataFrames
if i == 0:
extracted_features = pd.DataFrame([features_currentStudy])
else:
extracted_features = pd.concat( [extracted_features, pd.DataFrame([features_currentStudy])], ignore_index=True )
value_feature_names = [c_col for c_col in extracted_features.columns if (c_col.startswith('original') and '_shape_' not in c_col)]
dataset = extracted_features[value_feature_names]
# Saving files for later use
print("\nCreated dataset with shape:", dataset.shape, "with labels:", np.array(labels).shape)
dataset.to_pickle("image_dataset.pkl")
np.save("labels.npy", labels)
print("Saved dataset with filename: \"image_dataset.pkl\"")
# Extraction for external testing
import pathlib
test_path = "../testing_set/"
test_folder = pathlib.Path(test_path)
study_testset = list(test_folder.glob('P???.png'))
size_x, size_y = 128, 128
# Define tst_images array
tst_studies = np.zeros((100,size_x,size_y))
tst_masks = np.zeros((100,size_x,size_y))
for cname in (study_testset):
head_, cname_erase = ntpath.split(cname)
cname_erase = os.path.splitext(cname_erase)[0]
study_id = int(cname_erase[1:])
tst_imagePath = test_path + '/' + cname_erase + '.png'
tst_image = cv2.imread(tst_imagePath,cv2.IMREAD_GRAYSCALE)
tst_image = cv2.resize(tst_image,(size_x,size_y))
tst_image = np.array(tst_image)
tst_studies[study_id - 1] = tst_image
# Load corresponding mask
msk_imagePath = test_path + cname_erase + '_mask.png'
msk_image = cv2.imread(msk_imagePath,cv2.IMREAD_GRAYSCALE)
msk_image = cv2.resize(msk_image,(size_x,size_y))
msk_image = np.array(msk_image)
tst_masks[study_id - 1]+= np.array(msk_image)
tst_studies = tst_studies
tst_masks = tst_masks
tst_masks[tst_masks > 0] = 1
# Extract features from Benign and Malignant studies
for i in range(len(tst_studies)):
features_currentStudy = extractor.execute(GetImageFromArray(tst_studies[i]),
GetImageFromArray((tst_masks[i]).astype(np.uint8)),
label = 1)
# Stack DataFrames
if i == 0:
tst_extracted_features = pd.DataFrame([features_currentStudy])
else:
tst_extracted_features = pd.concat( [tst_extracted_features, pd.DataFrame([features_currentStudy])], ignore_index=True )
tst_extracted_features = tst_extracted_features[value_feature_names]
np_extst_features = tst_extracted_features.to_numpy()
# Save on file for reuse
tst_extracted_features.to_pickle("test.pkl")