Models:
Samuel Callahan
SCallahan/
Unsupervised_Anomaly
0
Downloads: 1
[978658]: / utils / default_config_setup.py

Download this file

272 lines (237 with data), 10.9 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
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
import json

import os

from enum import Enum



from dataloaders.BRAINWEB import BRAINWEB

from dataloaders.MSISBI2015 import MSISBI2015

from dataloaders.MSLUB import MSLUB

from dataloaders.MSSEG2008 import MSSEG2008



base_path = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))





class Dataset(Enum):

    BRAINWEB = 'BRAINWEBDIR'

    MSSEG2008_UNC = 'MSSEG2008DIR'

    MSSEG2008_CHB = 'MSSEG2008DIR'

    MSISBI2015 = 'MSISBI2015DIR'

    MSLUB = 'MSLUBDIR'





def get_options(batchsize, learningrate, numEpochs, zDim, outputWidth, outputHeight, slices_start=20, slices_end=130, numMonteCarloSamples=0, config=None):

    options = {}

    # Load config.json, which should hold DATADIR, CHECKPOINTDIR and SAMPLEDIR

    if config:

        options["globals"] = config

    else:

        with open(os.path.join(base_path, "config.default.json"), 'r') as f:

            options["globals"] = json.load(f)



    # Options

    options['debug'] = False

    options['data'] = {}

    options['train'] = {}

    options['train']['checkpointDir'] = options["globals"]["CHECKPOINTDIR"]

    options['train']['samplesDir'] = options["globals"]["SAMPLEDIR"]

    options['train']['batchsize'] = batchsize

    options['train']['learningrate'] = learningrate

    options['train']['numEpochs'] = numEpochs

    options['train']['zDim'] = zDim

    options['train']['snapshotAfter'] = 1000  # Take a snapshot after every 50 iterations

    options['train']['outputWidth'] = outputWidth

    options['train']['outputHeight'] = outputHeight

    options['train']['useTensorboard'] = True

    options['train']['useMatplotlib'] = False

    options['train']['tensorboardPort'] = 9001

    options['sliceStart'] = slices_start  # 20

    options['sliceEnd'] = slices_end  # 130

    options['threshold'] = 'bestdice'

    options['exportVolumes'] = False

    options['exportPRC'] = True

    options['exportROC'] = True

    options['numMonteCarloSamples'] = numMonteCarloSamples

    options['keepOnlyPositiveResiduals'] = True

    options['applyHyperIntensityPrior'] = True

    options['medianFiltering'] = True

    options['erodeBrainmask'] = True

    return options





def get_datasets(options, dataset: Dataset = Dataset.BRAINWEB):

    if dataset == Dataset.BRAINWEB:

        return get_Brainweb_healthy_dataset(options), get_Brainweb_lesion_dataset(options)

    elif dataset == Dataset.MSSEG2008_UNC:

        return None, get_MSSEG2008_dataset(options, 'UNC')

    elif dataset == Dataset.MSSEG2008_CHB:

        return None, get_MSSEG2008_dataset(options, 'CHB')

    elif dataset == Dataset.MSISBI2015:

        return None, get_MSISBI2015_dataset(options)

    elif dataset == Dataset.MSLUB:

        return None, get_MSLUB_dataset(options)

    else:

        raise ValueError(f'No valid dataset given: {dataset}')





###########################

#       MSSEG2008         #

###########################

def get_MSSEG2008_dataset(options, filter_sanner):

    dataset_options = get_MSSEG2008_dataset_options(options, filter_sanner)

    dataset = MSSEG2008(dataset_options)

    if options['debug']:

        dataset.visualize()



    return dataset





def get_MSSEG2008_dataset_options(options, filter_sanner):

    dataset_options = MSSEG2008.Options()

    dataset_options.description = ''

    dataset_options.debug = options['debug']

    dataset_options.dir = options['globals']['MSSEG2008DIR']

    dataset_options.useCrops = False

    dataset_options.cropType = 'center'  # Crop patches around lesions

    dataset_options.cropWidth = options['train']['outputWidth']

    dataset_options.cropHeight = options['train']['outputHeight']

    dataset_options.numRandomCropsPerSlice = 5  # Not needed when doing center crops

    dataset_options.rotations = [0]

    dataset_options.partition = {'TRAIN': 0, 'VAL': 2, 'TEST': 8}

    dataset_options.sliceResolution = [options['train']['outputHeight'], options['train']['outputWidth']]

    dataset_options.cache = True

    dataset_options.numSamples = -1

    dataset_options.addInstanceNoise = False

    dataset_options.axis = 'axial'

    dataset_options.filterScanner = filter_sanner  # 'UNC'or 'CHB'

    dataset_options.filterProtocols = ['FLAIR']

    dataset_options.filterType = "train"

    dataset_options.normalizationMethod = 'scaling'

    dataset_options.skullStripping = True

    dataset_options.sliceStart = options['sliceStart']

    dataset_options.sliceEnd = options['sliceEnd']

    dataset_options.skullStripping = True

    dataset_options.format = "aligned"



    return dataset_options





###########################

#       MSISBI2015        #

###########################

def get_MSISBI2015_dataset(options):

    dataset_options = get_MSISBI2015_dataset_options(options)

    dataset = MSISBI2015(dataset_options)

    if options['debug']:

        dataset.visualize()



    return dataset





def get_MSISBI2015_dataset_options(options):

    dataset_options = MSISBI2015.Options()

    dataset_options.description = ''

    dataset_options.debug = options['debug']

    dataset_options.dir = options['globals']['MSISBI2015DIR']

    dataset_options.useCrops = False

    dataset_options.cropType = 'center'  # Crop patches around lesions

    dataset_options.cropWidth = options['train']['outputWidth']

    dataset_options.cropHeight = options['train']['outputHeight']

    dataset_options.numRandomCropsPerSlice = 5  # Not needed when doing center crops

    dataset_options.rotations = [0]

    dataset_options.partition = {'TRAIN': 0, 'VAL': 5, 'TEST': 15}

    dataset_options.sliceResolution = [options['train']['outputHeight'], options['train']['outputWidth']]

    dataset_options.cache = True

    dataset_options.numSamples = -1

    dataset_options.addInstanceNoise = False

    dataset_options.axis = 'axial'

    dataset_options.filterProtocols = ['FLAIR']

    dataset_options.filterType = "train"

    dataset_options.normalizationMethod = 'scaling'

    dataset_options.skullStripping = True

    dataset_options.sliceStart = options['sliceStart']

    dataset_options.sliceEnd = options['sliceEnd']

    dataset_options.skullStripping = True

    dataset_options.format = "aligned"

    return dataset_options





###########################

#       MSLUB        #

###########################

def get_MSLUB_dataset(options):

    dataset_options = get_MSLUB_dataset_options(options)

    dataset = MSLUB(dataset_options)

    if options['debug']:

        dataset.visualize()



    return dataset





def get_MSLUB_dataset_options(options):

    dataset_options = MSLUB.Options()

    dataset_options.description = ''

    dataset_options.debug = options['debug']

    dataset_options.dir = options['globals']['MSLUBDIR']

    dataset_options.useCrops = False

    dataset_options.cropType = 'center'  # Crop patches around lesions

    dataset_options.cropWidth = options['train']['outputWidth']

    dataset_options.cropHeight = options['train']['outputHeight']

    dataset_options.numRandomCropsPerSlice = 5  # Not needed when doing center crops

    dataset_options.rotations = [0]

    dataset_options.partition = {'TRAIN': 0, 'VAL': 5, 'TEST': 25}

    dataset_options.sliceResolution = [options['train']['outputHeight'], options['train']['outputWidth']]

    dataset_options.cache = True

    dataset_options.numSamples = -1

    dataset_options.addInstanceNoise = False

    dataset_options.axis = 'axial'

    dataset_options.filterProtocols = ['FLAIR']

    dataset_options.normalizationMethod = 'scaling'

    dataset_options.skullStripping = True

    dataset_options.sliceStart = options['sliceStart']

    dataset_options.sliceEnd = options['sliceEnd']

    dataset_options.skullStripping = True

    dataset_options.format = "aligned"



    return dataset_options





#######################

#      Brainweb       #

#######################

def get_Brainweb_healthy_dataset(options):

    dataset_options = get_Brainweb_dataset_options(options)

    dataset_hc = BRAINWEB(dataset_options)

    if options['debug']:

        dataset_hc.visualize()

    return dataset_hc





def get_Brainweb_lesion_dataset(options):

    dataset_options = get_Brainweb_dataset_options(options)

    # Center Crops of slices from patients with lesions. Only for testing

    dataset_options.partition = {'TRAIN': 0.0, 'VAL': 0.0, 'TEST': 1.0}

    dataset_options.filterType = 'SEVEREMS'

    dataset_options.rotations = [0]

    return BRAINWEB(dataset_options)





def get_Brainweb_dataset_options(options):

    dataset_options = BRAINWEB.Options()

    dataset_options.description = ""

    dataset_options.debug = options['debug']

    dataset_options.dir = options['data']['dir']

    dataset_options.useCrops = False

    dataset_options.cropType = 'center'  # Not used when useCrops is False

    dataset_options.cropWidth = options['train']['outputWidth']

    dataset_options.cropHeight = options['train']['outputHeight']

    dataset_options.numRandomCropsPerSlice = 5  # Not needed when doing center crops

    dataset_options.rotations = [0]

    dataset_options.partition = {'TRAIN': 0.7, 'VAL': 0.3, 'TEST': 0.0}

    dataset_options.sliceResolution = [options['train']['outputHeight'], options['train']['outputWidth']]

    dataset_options.cache = True

    dataset_options.numSamples = -1

    dataset_options.addInstanceNoise = False

    dataset_options.axis = 'axial'

    dataset_options.filterType = 'NORMAL'

    dataset_options.filterProtocol = 'T2'

    dataset_options.normalizationMethod = 'scaling'

    dataset_options.skullRemoval = True

    dataset_options.sliceStart = options['sliceStart']

    dataset_options.sliceEnd = options['sliceEnd']

    dataset_options.backgroundRemoval = True

    dataset_options.registerTo = None

    return dataset_options





def get_config(trainer, options, optimizer, intermediateResolutions, dropout_rate, dataset):

    config = trainer.Config()

    config.dataset = type(dataset).__name__

    config.description = ''

    config.numChannels = dataset.num_channels

    config.batchsize = options['train']['batchsize']

    config.checkpointDir = options['train']['checkpointDir']

    config.snapShotAfter = options['train']['snapshotAfter']

    config.sampleDir = options['train']['samplesDir']

    config.learningrate = options['train']['learningrate']

    config.numEpochs = options['train']['numEpochs']

    config.zDim = options['train']['zDim']

    config.beta1 = 0.5

    config.outputHeight = options['train']['outputHeight']

    config.outputWidth = options['train']['outputWidth']

    config.useTensorboard = options['train']['useTensorboard']

    config.useMatplotlib = options['train']['useMatplotlib']

    config.tensorboardPort = options['train']['tensorboardPort']

    config.debugGradients = options['debug']

    config.optimizer = optimizer

    config.intermediateResolutions = intermediateResolutions

    config.weightRegularization = 0.0

    config.dropout_rate = dropout_rate

    config.dropout = False

    config.l1_weight = 1.0

    config.options = options

    return config