-
Notifications
You must be signed in to change notification settings - Fork 4
/
build_chaos_dataset.py
198 lines (156 loc) · 6.52 KB
/
build_chaos_dataset.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
193
194
195
196
# encoding: utf-8
__author__ = 'Jonas Teuwen'
import os
import sys
import re
import numpy as np
import PIL.Image
import argparse
from tqdm import tqdm
from glob import glob
import SimpleITK as sitk
from collections import defaultdict
# Mapping
# 0 is background
# 1 is liver
# 2 is right kidney
# 3 is left kidney
# 4 is spleen
def class_mapping(input_value):
if 55 < input_value <= 70:
return 1
elif 110 < input_value <= 135:
return 2
elif 175 < input_value <= 200:
return 3
elif 240 < input_value <= 255:
return 4
else:
return 0
def parse_args():
"""Parse input arguments"""
parser = argparse.ArgumentParser(description='Parse CHAOS dataset')
parser.add_argument(
'modality',
help='modality, either MR or CT',)
parser.add_argument(
'root_dir',
help='root to data',)
parser.add_argument(
'write_to',
help='folder to write output to', )
return parser.parse_args()
def get_patients(path):
patients = []
regex = '^\d+$'
for x in os.listdir(path):
if re.match(regex, x):
patients.append(x)
return patients
def get_masks(gt_images, vol_img, mask_name):
all_masks = []
# I need to do this for the CT masks, the seem to be flipped.
if mask_name == 'liver':
gt_images = gt_images[::-1]
gt_arr = np.stack([np.asarray(PIL.Image.open(_)) for _ in gt_images])
unique_values_mask = np.unique(gt_arr)
gt_mask = np.zeros_like(gt_arr).astype(np.uint8)
for unique_value in unique_values_mask:
if mask_name == 'liver':
gt_mask[gt_arr.astype(np.int) == 1] = 1
else:
gt_mask[gt_arr == unique_value] = class_mapping(unique_value)
gt_sitk_mask = sitk.GetImageFromArray(gt_mask)
gt_sitk_mask.SetOrigin(vol_img.GetOrigin())
gt_sitk_mask.SetDirection(vol_img.GetDirection())
gt_sitk_mask.SetSpacing(vol_img.GetSpacing())
all_masks.append((mask_name, gt_sitk_mask))
return all_masks
def get_mri_images_from_patient(patient_path):
all_images = []
for sequence_type in ['T1DUAL', 'T2SPIR']:
dicoms = os.path.join(patient_path, sequence_type, 'DICOM_anon')
dcm_images = glob(os.path.join(dicoms, '**', 'IMG*.dcm'), recursive=True)
gt = os.path.join(patient_path, sequence_type, 'Ground')
gt_images = glob(os.path.join(gt, '*'))
gt_images = sorted(gt_images, key=lambda x: int(os.path.basename(x).split('-')[-1].split('.png')[0]))
# Try to read the dicom images
slice_thicknesses = []
images_dict = defaultdict(list)
location_dict = {}
for dcm in dcm_images:
file_reader = sitk.ImageFileReader()
file_reader.SetFileName(dcm)
file_reader.ReadImageInformation()
slice_thickness = float(file_reader.GetMetaData('0018|0050').strip())
slice_thicknesses.append(slice_thickness)
slice_location = float(file_reader.GetMetaData('0020|1041').strip())
echo_time = float(file_reader.GetMetaData('0018|0081'))
images_dict[echo_time].append(dcm)
location_dict[dcm] = slice_location
assert len(set(slice_thicknesses)) == 1, f'Multiple thicknesses in images: {slice_thicknesses}'
for echo_time, dcm_fns in images_dict.items():
dcm_fns = sorted(dcm_fns, key=lambda x: location_dict[x])
slices = [sitk.ReadImage(_) for _ in dcm_fns]
vol_img = sitk.Image(slices[0].GetSize()[0], slices[0].GetSize()[1], len(slices), slices[0].GetPixelID())
for idx_z, slice_vol in enumerate(slices):
vol_img = sitk.Paste(vol_img, slice_vol, slice_vol.GetSize(), destinationIndex=[0, 0, idx_z])
vol_img.SetSpacing(slices[0].GetSpacing())
vol_img.SetOrigin(slices[0].GetOrigin())
all_images.append((sequence_type, echo_time, vol_img))
all_masks = get_masks(gt_images, vol_img, sequence_type)
return all_images, all_masks
def get_ct_images_from_patient(patient_path):
dicoms = os.path.join(patient_path, 'DICOM_anon')
gt = os.path.join(patient_path, 'Ground')
gt_images = glob(os.path.join(gt, '*.png'))
gt_images = sorted(gt_images, key=lambda x: int(os.path.basename(x).split('_')[-1].split('.png')[0]))
# Try to read the dicom images
reader = sitk.ImageSeriesReader()
series_ids = list(reader.GetGDCMSeriesIDs(dicoms))
assert len(series_ids) == 1, 'Assuming one series id'
fns = reader.GetGDCMSeriesFileNames(dicoms, series_ids[0])
reader.SetFileNames(fns)
vol_img = reader.Execute()
all_masks = get_masks(gt_images, vol_img, 'liver')
return vol_img, all_masks
def main_mri(args):
patients = get_patients(args.root_dir)
for patient in tqdm(patients):
images, masks = get_mri_images_from_patient(os.path.join(args.root_dir, patient))
images.sort(key=lambda x: x[1]) # Sort on echo time, longer echo time is the in-phase image
for idx, image_list in enumerate(images):
sequence_type, echo_time, image = image_list
if sequence_type == 'T1DUAL':
if idx == 0:
fn = f'T1DUAL_out_phase_image.nrrd'
else:
fn = f'T1DUAL_in_phase_image.nrrd'
elif sequence_type == 'T2SPIR':
fn = 'T2SPIR_image.nrrd'
write_to_folder = os.path.join(args.write_to, f'Patient_{patient}')
os.makedirs(write_to_folder, exist_ok=True)
sitk.WriteImage(image, os.path.join(write_to_folder, fn), True)
for mask_name, mask in masks:
fn = f'{mask_name}_mask.nrrd'
write_to_folder = os.path.join(args.write_to, f'Patient_{patient}')
sitk.WriteImage(mask, os.path.join(write_to_folder, fn), True)
def main_ct(args):
patients = get_patients(args.root_dir)
for patient in tqdm(patients):
image, masks = get_ct_images_from_patient(os.path.join(args.root_dir, patient))
mask_name, mask = masks[0]
write_to_folder = os.path.join(args.write_to, f'Patient_{patient}')
os.makedirs(write_to_folder, exist_ok=True)
sitk.WriteImage(image, os.path.join(write_to_folder, 'CT_image.nrrd'), True)
sitk.WriteImage(mask, os.path.join(write_to_folder, f'{mask_name}_mask.nrrd'), True)
def main():
args = parse_args()
if args.modality == 'CT':
main_ct(args)
elif args.modality == 'MR':
main_mri(args)
else:
sys.exit('Choose MR or CT as modality.')
if __name__ == '__main__':
main()