File size: 20,639 Bytes
ecf08bc |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 |
# Copyright 2020 Division of Medical Image Computing, German Cancer Research Center (DKFZ), Heidelberg, Germany
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from PIL import Image
import shutil
from collections import OrderedDict
import dicom2nifti
import numpy as np
from batchgenerators.utilities.data_splitting import get_split_deterministic
from batchgenerators.utilities.file_and_folder_operations import *
from PIL import Image
import SimpleITK as sitk
from nnunet.paths import preprocessing_output_dir, nnUNet_raw_data
from nnunet.utilities.sitk_stuff import copy_geometry
from nnunet.inference.ensemble_predictions import merge
def load_png_stack(folder):
pngs = subfiles(folder, suffix="png")
pngs.sort()
loaded = []
for p in pngs:
loaded.append(np.array(Image.open(p)))
loaded = np.stack(loaded, 0)[::-1]
return loaded
def convert_CT_seg(loaded_png):
return loaded_png.astype(np.uint16)
def convert_MR_seg(loaded_png):
result = np.zeros(loaded_png.shape)
result[(loaded_png > 55) & (loaded_png <= 70)] = 1 # liver
result[(loaded_png > 110) & (loaded_png <= 135)] = 2 # right kidney
result[(loaded_png > 175) & (loaded_png <= 200)] = 3 # left kidney
result[(loaded_png > 240) & (loaded_png <= 255)] = 4 # spleen
return result
def convert_seg_to_intensity_task5(seg):
seg_new = np.zeros(seg.shape, dtype=np.uint8)
seg_new[seg == 1] = 63
seg_new[seg == 2] = 126
seg_new[seg == 3] = 189
seg_new[seg == 4] = 252
return seg_new
def convert_seg_to_intensity_task3(seg):
seg_new = np.zeros(seg.shape, dtype=np.uint8)
seg_new[seg == 1] = 63
return seg_new
def write_pngs_from_nifti(nifti, output_folder, converter=convert_seg_to_intensity_task3):
npy = sitk.GetArrayFromImage(sitk.ReadImage(nifti))
seg_new = converter(npy)
for z in range(len(npy)):
Image.fromarray(seg_new[z]).save(join(output_folder, "img%03.0d.png" % z))
def convert_variant2_predicted_test_to_submission_format(folder_with_predictions,
output_folder="/home/fabian/drives/datasets/results/nnUNet/test_sets/Task038_CHAOS_Task_3_5_Variant2/ready_to_submit",
postprocessing_file="/home/fabian/drives/datasets/results/nnUNet/ensembles/Task038_CHAOS_Task_3_5_Variant2/ensemble_2d__nnUNetTrainerV2__nnUNetPlansv2.1--3d_fullres__nnUNetTrainerV2__nnUNetPlansv2.1/postprocessing.json"):
"""
output_folder is where the extracted template is
:param folder_with_predictions:
:param output_folder:
:return:
"""
postprocessing_file = "/media/fabian/Results/nnUNet/3d_fullres/Task039_CHAOS_Task_3_5_Variant2_highres/" \
"nnUNetTrainerV2__nnUNetPlansfixed/postprocessing.json"
# variant 2 treats in and out phase as two training examples, so we need to ensemble these two again
final_predictions_folder = join(output_folder, "final")
maybe_mkdir_p(final_predictions_folder)
t1_patient_names = [i.split("_")[-1][:-7] for i in subfiles(folder_with_predictions, prefix="T1", suffix=".nii.gz", join=False)]
folder_for_ensembing0 = join(output_folder, "ens0")
folder_for_ensembing1 = join(output_folder, "ens1")
maybe_mkdir_p(folder_for_ensembing0)
maybe_mkdir_p(folder_for_ensembing1)
# now copy all t1 out phases in ens0 and all in phases in ens1. Name them the same.
for t1 in t1_patient_names:
shutil.copy(join(folder_with_predictions, "T1_in_%s.npz" % t1), join(folder_for_ensembing1, "T1_%s.npz" % t1))
shutil.copy(join(folder_with_predictions, "T1_in_%s.pkl" % t1), join(folder_for_ensembing1, "T1_%s.pkl" % t1))
shutil.copy(join(folder_with_predictions, "T1_out_%s.npz" % t1), join(folder_for_ensembing0, "T1_%s.npz" % t1))
shutil.copy(join(folder_with_predictions, "T1_out_%s.pkl" % t1), join(folder_for_ensembing0, "T1_%s.pkl" % t1))
shutil.copy(join(folder_with_predictions, "plans.pkl"), join(folder_for_ensembing0, "plans.pkl"))
shutil.copy(join(folder_with_predictions, "plans.pkl"), join(folder_for_ensembing1, "plans.pkl"))
# there is a problem with T1_35 that I need to correct manually (different crop size, will not negatively impact results)
#ens0_softmax = np.load(join(folder_for_ensembing0, "T1_35.npz"))['softmax']
ens1_softmax = np.load(join(folder_for_ensembing1, "T1_35.npz"))['softmax']
#ens0_props = load_pickle(join(folder_for_ensembing0, "T1_35.pkl"))
#ens1_props = load_pickle(join(folder_for_ensembing1, "T1_35.pkl"))
ens1_softmax = ens1_softmax[:, :, :-1, :]
np.savez_compressed(join(folder_for_ensembing1, "T1_35.npz"), softmax=ens1_softmax)
shutil.copy(join(folder_for_ensembing0, "T1_35.pkl"), join(folder_for_ensembing1, "T1_35.pkl"))
# now call my ensemble function
merge((folder_for_ensembing0, folder_for_ensembing1), final_predictions_folder, 8, True,
postprocessing_file=postprocessing_file)
# copy t2 files to final_predictions_folder as well
t2_files = subfiles(folder_with_predictions, prefix="T2", suffix=".nii.gz", join=False)
for t2 in t2_files:
shutil.copy(join(folder_with_predictions, t2), join(final_predictions_folder, t2))
# apply postprocessing
from nnunet.postprocessing.connected_components import apply_postprocessing_to_folder, load_postprocessing
postprocessed_folder = join(output_folder, "final_postprocessed")
for_which_classes, min_valid_obj_size = load_postprocessing(postprocessing_file)
apply_postprocessing_to_folder(final_predictions_folder, postprocessed_folder,
for_which_classes, min_valid_obj_size, 8)
# now export the niftis in the weird png format
# task 3
output_dir = join(output_folder, "CHAOS_submission_template_new", "Task3", "MR")
for t1 in t1_patient_names:
output_folder_here = join(output_dir, t1, "T1DUAL", "Results")
nifti_file = join(postprocessed_folder, "T1_%s.nii.gz" % t1)
write_pngs_from_nifti(nifti_file, output_folder_here, converter=convert_seg_to_intensity_task3)
for t2 in t2_files:
patname = t2.split("_")[-1][:-7]
output_folder_here = join(output_dir, patname, "T2SPIR", "Results")
nifti_file = join(postprocessed_folder, "T2_%s.nii.gz" % patname)
write_pngs_from_nifti(nifti_file, output_folder_here, converter=convert_seg_to_intensity_task3)
# task 5
output_dir = join(output_folder, "CHAOS_submission_template_new", "Task5", "MR")
for t1 in t1_patient_names:
output_folder_here = join(output_dir, t1, "T1DUAL", "Results")
nifti_file = join(postprocessed_folder, "T1_%s.nii.gz" % t1)
write_pngs_from_nifti(nifti_file, output_folder_here, converter=convert_seg_to_intensity_task5)
for t2 in t2_files:
patname = t2.split("_")[-1][:-7]
output_folder_here = join(output_dir, patname, "T2SPIR", "Results")
nifti_file = join(postprocessed_folder, "T2_%s.nii.gz" % patname)
write_pngs_from_nifti(nifti_file, output_folder_here, converter=convert_seg_to_intensity_task5)
if __name__ == "__main__":
"""
This script only prepares data to participate in Task 5 and Task 5. I don't like the CT task because
1) there are
no abdominal organs in the ground truth. In the case of CT we are supposed to train only liver while on MRI we are
supposed to train all organs. This would require manual modification of nnU-net to deal with this dataset. This is
not what nnU-net is about.
2) CT Liver or multiorgan segmentation is too easy to get external data for. Therefore the challenges comes down
to who gets the b est external data, not who has the best algorithm. Not super interesting.
Task 3 is a subtask of Task 5 so we need to prepare the data only once.
Difficulty: We need to process both T1 and T2, but T1 has 2 'modalities' (phases). nnU-Net cannot handly varying
number of input channels. We need to be creative.
We deal with this by preparing 2 Variants:
1) pretend we have 2 modalities for T2 as well by simply stacking a copy of the data
2) treat all MRI sequences independently, so we now have 3*20 training data instead of 2*20. In inference we then
ensemble the results for the two t1 modalities.
Careful: We need to split manually here to ensure we stratify by patient
"""
root = "/media/fabian/My Book/datasets/CHAOS_challenge/Train_Sets"
root_test = "/media/fabian/My Book/datasets/CHAOS_challenge/Test_Sets"
out_base = nnUNet_raw_data
# CT
# we ignore CT because
##############################################################
# Variant 1
##############################################################
patient_ids = []
patient_ids_test = []
output_folder = join(out_base, "Task037_CHAOS_Task_3_5_Variant1")
output_images = join(output_folder, "imagesTr")
output_labels = join(output_folder, "labelsTr")
output_imagesTs = join(output_folder, "imagesTs")
maybe_mkdir_p(output_images)
maybe_mkdir_p(output_labels)
maybe_mkdir_p(output_imagesTs)
# Process T1 train
d = join(root, "MR")
patients = subdirs(d, join=False)
for p in patients:
patient_name = "T1_" + p
gt_dir = join(d, p, "T1DUAL", "Ground")
seg = convert_MR_seg(load_png_stack(gt_dir)[::-1])
img_dir = join(d, p, "T1DUAL", "DICOM_anon", "InPhase")
img_outfile = join(output_images, patient_name + "_0000.nii.gz")
_ = dicom2nifti.convert_dicom.dicom_series_to_nifti(img_dir, img_outfile, reorient_nifti=False)
img_dir = join(d, p, "T1DUAL", "DICOM_anon", "OutPhase")
img_outfile = join(output_images, patient_name + "_0001.nii.gz")
_ = dicom2nifti.convert_dicom.dicom_series_to_nifti(img_dir, img_outfile, reorient_nifti=False)
img_sitk = sitk.ReadImage(img_outfile)
img_sitk_npy = sitk.GetArrayFromImage(img_sitk)
seg_itk = sitk.GetImageFromArray(seg.astype(np.uint8))
seg_itk = copy_geometry(seg_itk, img_sitk)
sitk.WriteImage(seg_itk, join(output_labels, patient_name + ".nii.gz"))
patient_ids.append(patient_name)
# Process T1 test
d = join(root_test, "MR")
patients = subdirs(d, join=False)
for p in patients:
patient_name = "T1_" + p
img_dir = join(d, p, "T1DUAL", "DICOM_anon", "InPhase")
img_outfile = join(output_imagesTs, patient_name + "_0000.nii.gz")
_ = dicom2nifti.convert_dicom.dicom_series_to_nifti(img_dir, img_outfile, reorient_nifti=False)
img_dir = join(d, p, "T1DUAL", "DICOM_anon", "OutPhase")
img_outfile = join(output_imagesTs, patient_name + "_0001.nii.gz")
_ = dicom2nifti.convert_dicom.dicom_series_to_nifti(img_dir, img_outfile, reorient_nifti=False)
img_sitk = sitk.ReadImage(img_outfile)
img_sitk_npy = sitk.GetArrayFromImage(img_sitk)
patient_ids_test.append(patient_name)
# Process T2 train
d = join(root, "MR")
patients = subdirs(d, join=False)
for p in patients:
patient_name = "T2_" + p
gt_dir = join(d, p, "T2SPIR", "Ground")
seg = convert_MR_seg(load_png_stack(gt_dir)[::-1])
img_dir = join(d, p, "T2SPIR", "DICOM_anon")
img_outfile = join(output_images, patient_name + "_0000.nii.gz")
_ = dicom2nifti.convert_dicom.dicom_series_to_nifti(img_dir, img_outfile, reorient_nifti=False)
shutil.copy(join(output_images, patient_name + "_0000.nii.gz"), join(output_images, patient_name + "_0001.nii.gz"))
img_sitk = sitk.ReadImage(img_outfile)
img_sitk_npy = sitk.GetArrayFromImage(img_sitk)
seg_itk = sitk.GetImageFromArray(seg.astype(np.uint8))
seg_itk = copy_geometry(seg_itk, img_sitk)
sitk.WriteImage(seg_itk, join(output_labels, patient_name + ".nii.gz"))
patient_ids.append(patient_name)
# Process T2 test
d = join(root_test, "MR")
patients = subdirs(d, join=False)
for p in patients:
patient_name = "T2_" + p
gt_dir = join(d, p, "T2SPIR", "Ground")
img_dir = join(d, p, "T2SPIR", "DICOM_anon")
img_outfile = join(output_imagesTs, patient_name + "_0000.nii.gz")
_ = dicom2nifti.convert_dicom.dicom_series_to_nifti(img_dir, img_outfile, reorient_nifti=False)
shutil.copy(join(output_imagesTs, patient_name + "_0000.nii.gz"), join(output_imagesTs, patient_name + "_0001.nii.gz"))
img_sitk = sitk.ReadImage(img_outfile)
img_sitk_npy = sitk.GetArrayFromImage(img_sitk)
patient_ids_test.append(patient_name)
json_dict = OrderedDict()
json_dict['name'] = "Chaos Challenge Task3/5 Variant 1"
json_dict['description'] = "nothing"
json_dict['tensorImageSize'] = "4D"
json_dict['reference'] = "https://chaos.grand-challenge.org/Data/"
json_dict['licence'] = "see https://chaos.grand-challenge.org/Data/"
json_dict['release'] = "0.0"
json_dict['modality'] = {
"0": "MRI",
"1": "MRI",
}
json_dict['labels'] = {
"0": "background",
"1": "liver",
"2": "right kidney",
"3": "left kidney",
"4": "spleen",
}
json_dict['numTraining'] = len(patient_ids)
json_dict['numTest'] = 0
json_dict['training'] = [{'image': "./imagesTr/%s.nii.gz" % i, "label": "./labelsTr/%s.nii.gz" % i} for i in
patient_ids]
json_dict['test'] = []
save_json(json_dict, join(output_folder, "dataset.json"))
##############################################################
# Variant 2
##############################################################
patient_ids = []
patient_ids_test = []
output_folder = join(out_base, "Task038_CHAOS_Task_3_5_Variant2")
output_images = join(output_folder, "imagesTr")
output_imagesTs = join(output_folder, "imagesTs")
output_labels = join(output_folder, "labelsTr")
maybe_mkdir_p(output_images)
maybe_mkdir_p(output_imagesTs)
maybe_mkdir_p(output_labels)
# Process T1 train
d = join(root, "MR")
patients = subdirs(d, join=False)
for p in patients:
patient_name_in = "T1_in_" + p
patient_name_out = "T1_out_" + p
gt_dir = join(d, p, "T1DUAL", "Ground")
seg = convert_MR_seg(load_png_stack(gt_dir)[::-1])
img_dir = join(d, p, "T1DUAL", "DICOM_anon", "InPhase")
img_outfile = join(output_images, patient_name_in + "_0000.nii.gz")
_ = dicom2nifti.convert_dicom.dicom_series_to_nifti(img_dir, img_outfile, reorient_nifti=False)
img_dir = join(d, p, "T1DUAL", "DICOM_anon", "OutPhase")
img_outfile = join(output_images, patient_name_out + "_0000.nii.gz")
_ = dicom2nifti.convert_dicom.dicom_series_to_nifti(img_dir, img_outfile, reorient_nifti=False)
img_sitk = sitk.ReadImage(img_outfile)
img_sitk_npy = sitk.GetArrayFromImage(img_sitk)
seg_itk = sitk.GetImageFromArray(seg.astype(np.uint8))
seg_itk = copy_geometry(seg_itk, img_sitk)
sitk.WriteImage(seg_itk, join(output_labels, patient_name_in + ".nii.gz"))
sitk.WriteImage(seg_itk, join(output_labels, patient_name_out + ".nii.gz"))
patient_ids.append(patient_name_out)
patient_ids.append(patient_name_in)
# Process T1 test
d = join(root_test, "MR")
patients = subdirs(d, join=False)
for p in patients:
patient_name_in = "T1_in_" + p
patient_name_out = "T1_out_" + p
gt_dir = join(d, p, "T1DUAL", "Ground")
img_dir = join(d, p, "T1DUAL", "DICOM_anon", "InPhase")
img_outfile = join(output_imagesTs, patient_name_in + "_0000.nii.gz")
_ = dicom2nifti.convert_dicom.dicom_series_to_nifti(img_dir, img_outfile, reorient_nifti=False)
img_dir = join(d, p, "T1DUAL", "DICOM_anon", "OutPhase")
img_outfile = join(output_imagesTs, patient_name_out + "_0000.nii.gz")
_ = dicom2nifti.convert_dicom.dicom_series_to_nifti(img_dir, img_outfile, reorient_nifti=False)
img_sitk = sitk.ReadImage(img_outfile)
img_sitk_npy = sitk.GetArrayFromImage(img_sitk)
patient_ids_test.append(patient_name_out)
patient_ids_test.append(patient_name_in)
# Process T2 train
d = join(root, "MR")
patients = subdirs(d, join=False)
for p in patients:
patient_name = "T2_" + p
gt_dir = join(d, p, "T2SPIR", "Ground")
seg = convert_MR_seg(load_png_stack(gt_dir)[::-1])
img_dir = join(d, p, "T2SPIR", "DICOM_anon")
img_outfile = join(output_images, patient_name + "_0000.nii.gz")
_ = dicom2nifti.convert_dicom.dicom_series_to_nifti(img_dir, img_outfile, reorient_nifti=False)
img_sitk = sitk.ReadImage(img_outfile)
img_sitk_npy = sitk.GetArrayFromImage(img_sitk)
seg_itk = sitk.GetImageFromArray(seg.astype(np.uint8))
seg_itk = copy_geometry(seg_itk, img_sitk)
sitk.WriteImage(seg_itk, join(output_labels, patient_name + ".nii.gz"))
patient_ids.append(patient_name)
# Process T2 test
d = join(root_test, "MR")
patients = subdirs(d, join=False)
for p in patients:
patient_name = "T2_" + p
gt_dir = join(d, p, "T2SPIR", "Ground")
img_dir = join(d, p, "T2SPIR", "DICOM_anon")
img_outfile = join(output_imagesTs, patient_name + "_0000.nii.gz")
_ = dicom2nifti.convert_dicom.dicom_series_to_nifti(img_dir, img_outfile, reorient_nifti=False)
img_sitk = sitk.ReadImage(img_outfile)
img_sitk_npy = sitk.GetArrayFromImage(img_sitk)
patient_ids_test.append(patient_name)
json_dict = OrderedDict()
json_dict['name'] = "Chaos Challenge Task3/5 Variant 2"
json_dict['description'] = "nothing"
json_dict['tensorImageSize'] = "4D"
json_dict['reference'] = "https://chaos.grand-challenge.org/Data/"
json_dict['licence'] = "see https://chaos.grand-challenge.org/Data/"
json_dict['release'] = "0.0"
json_dict['modality'] = {
"0": "MRI",
}
json_dict['labels'] = {
"0": "background",
"1": "liver",
"2": "right kidney",
"3": "left kidney",
"4": "spleen",
}
json_dict['numTraining'] = len(patient_ids)
json_dict['numTest'] = 0
json_dict['training'] = [{'image': "./imagesTr/%s.nii.gz" % i, "label": "./labelsTr/%s.nii.gz" % i} for i in
patient_ids]
json_dict['test'] = []
save_json(json_dict, join(output_folder, "dataset.json"))
#################################################
# custom split
#################################################
patients = subdirs(join(root, "MR"), join=False)
task_name_variant1 = "Task037_CHAOS_Task_3_5_Variant1"
task_name_variant2 = "Task038_CHAOS_Task_3_5_Variant2"
output_preprocessed_v1 = join(preprocessing_output_dir, task_name_variant1)
maybe_mkdir_p(output_preprocessed_v1)
output_preprocessed_v2 = join(preprocessing_output_dir, task_name_variant2)
maybe_mkdir_p(output_preprocessed_v2)
splits = []
for fold in range(5):
tr, val = get_split_deterministic(patients, fold, 5, 12345)
train = ["T2_" + i for i in tr] + ["T1_" + i for i in tr]
validation = ["T2_" + i for i in val] + ["T1_" + i for i in val]
splits.append({
'train': train,
'val': validation
})
save_pickle(splits, join(output_preprocessed_v1, "splits_final.pkl"))
splits = []
for fold in range(5):
tr, val = get_split_deterministic(patients, fold, 5, 12345)
train = ["T2_" + i for i in tr] + ["T1_in_" + i for i in tr] + ["T1_out_" + i for i in tr]
validation = ["T2_" + i for i in val] + ["T1_in_" + i for i in val] + ["T1_out_" + i for i in val]
splits.append({
'train': train,
'val': validation
})
save_pickle(splits, join(output_preprocessed_v2, "splits_final.pkl"))
|