BrainMapperUNet.py 18 KB
Newer Older
1
2
3
4
"""Brain Mapper U-Net Architecture

Description:

5
6
    This folder contains the Pytorch implementation of the core U-net architecture.
    This arcitecture predicts functional connectivity rsfMRI from structural connectivity information from dMRI.
7

8
9
10
11
12
13
Usage:

    To use this module, import it and instantiate is as you wish:

        from BrainMapperUNet import BrainMapperUNet
        deep_learning_model = BrainMapperUnet(parameters)
14
15
16
17
18
19

"""

import numpy as np
import torch
import torch.nn as nn
20
import utils.modules as modules
21

22

23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
class BrainMapperUNet(nn.Module):
    """Architecture class BrainMapper U-net.

    This class contains the pytorch implementation of the U-net architecture underpinning the BrainMapper project.

    Args:
        parameters (dict): Contains information relevant parameters
        parameters = {
            'kernel_heigth': 5
            'kernel_width': 5
            'kernel_classification': 1
            'input_channels': 1
            'output_channels': 64
            'convolution_stride': 1
            'dropout': 0.2
            'pool_kernel_size': 2
            'pool_stride': 2
            'up_mode': 'upconv'
            'number_of_classes': 1
        }
43

44
45
    Returns:
        probability_map (torch.tensor): Output forward passed tensor through the U-net block
46
    """
47

48
    def __init__(self, parameters):
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
        super(BrainMapperUNet, self).__init__()

        # TODO: currently, architecture based on QuickNAT - need to adjust parameter values accordingly!

        self.encoderBlock1 = modules.EncoderBlock(parameters)
        parameters['input_channels'] = parameters['output_channels']
        self.encoderBlock2 = modules.EncoderBlock(parameters)
        self.encoderBlock3 = modules.EncoderBlock(parameters)
        self.encoderBlock4 = modules.EncoderBlock(parameters)

        self.bottleneck = modules.ConvolutionalBlock(parameters)

        parameters['input_channels'] = parameters['output_channels'] * 2.0
        self.decoderBlock1 = modules.DecoderBlock(parameters)
        self.decoderBlock2 = modules.DecoderBlock(parameters)
        self.decoderBlock3 = modules.DecoderBlock(parameters)
        self.decoderBlock4 = modules.DecoderBlock(parameters)

        parameters['input_channels'] = parameters['output_channels']
        self.classifier = modules.ClassifierBlock(parameters)
69
70

    def forward(self, X):
71
72
73
74
75
76
77
78
79
80
        """Forward pass for U-net

        Function computing the forward pass through the U-Net
        The input to the function is the dMRI map

        Args:
            X (torch.tensor): Input dMRI map, shape = (N x C x H x W) 

        Returns:
            probability_map (torch.tensor): Output forward passed tensor through the U-net block
81
        """
82

83
84
85
        Y_encoder_1, Y_np1, pool_indices1 = self.encoderBlock1.forward(X)
        Y_encoder_2, Y_np2, pool_indices2 = self.encoderBlock2.forward(
            Y_encoder_1)
86
87
88

        del Y_encoder_1

89
90
        Y_encoder_3, Y_np3, pool_indices3 = self.encoderBlock3.forward(
            Y_encoder_2)
91
92
93

        del Y_encoder_2

94
95
96
        Y_encoder_4, Y_np4, pool_indices4 = self.encoderBlock4.forward(
            Y_encoder_3)

97
98
        del Y_encoder_3

99
100
        Y_bottleNeck = self.bottleneck.forward(Y_encoder_4)

101
102
        del Y_encoder_4

103
104
        Y_decoder_1 = self.decoderBlock1.forward(
            Y_bottleNeck, Y_np4, pool_indices4)
105
106
107

        del Y_bottleNeck, Y_np4, pool_indices4
        
108
109
        Y_decoder_2 = self.decoderBlock2.forward(
            Y_decoder_1, Y_np3, pool_indices3)
110
111
112

        del Y_decoder_1, Y_np3, pool_indices3

113
114
        Y_decoder_3 = self.decoderBlock3.forward(
            Y_decoder_2, Y_np2, pool_indices2)
115
116
117

        del Y_decoder_2, Y_np2, pool_indices2

118
        Y_decoder_4 = self.decoderBlock4.forward(
119
120
            Y_decoder_3, Y_np1, pool_indices1)

121
122
        del Y_decoder_3, Y_np1, pool_indices1

123
124
        probability_map = self.classifier.forward(Y_decoder_4)

125
126
        del Y_decoder_4

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
        return probability_map

    def save(self, path):
        """Model Saver

        Function saving the model with all its parameters to a given path.
        The path must end with a *.model argument.

        Args:
            path (str): Path string
        """

        print("Saving Model... {}".format(path))
        torch.save(self, path)

    @property
    def test_if_cuda(self):
        """Cuda Test

        This function tests if the model parameters are allocated to a CUDA enabled GPU.

        Returns:
            bool: Flag indicating True if the tensor is stored on the GPU and Flase otherwhise
        """

        return next(self.parameters()).is_cuda

    def predict(self, X, device=0):
        """Post-training Output Prediction

        This function predicts the output of the of the U-net post-training

        Args:
            X (torch.tensor): input dMRI volume
            device (int/str): Device type used for training (int - GPU id, str- CPU)

        Returns:
            prediction (ndarray): predicted output after training

        """
        self.eval()  # PyToch module setting network to evaluation mode

        if type(X) is np.ndarray:
            X = torch.tensor(X, requires_grad=False).type(torch.FloatTensor)
        elif type(X) is torch.Tensor and not X.is_cuda:
            X = X.type(torch.FloatTensor).cuda(device, non_blocking=True)

        # .cuda() call transfers the densor from the CPU to the GPU if that is the case.
        # Non-blocking argument lets the caller bypas synchronization when necessary

        with torch.no_grad():  # Causes operations to have no gradients
            output = self.forward(X)

        _, idx = torch.max(output, 1)

        # We retrieve the tensor held by idx (.data), and map it to a cpu as an ndarray
        idx = idx.data.cpu().numpy()

        prediction = np.squeeze(idx)

        del X, output, idx

        return prediction

class BrainMapperUNet3D(nn.Module):
    """Architecture class BrainMapper 3D U-net.

    This class contains the pytorch implementation of the U-net architecture underpinning the BrainMapper project.

    Args:
        parameters (dict): Contains information relevant parameters
        parameters = {
            'kernel_heigth': 5
            'kernel_width': 5
            'kernel_depth': 5
            'kernel_classification': 1
            'input_channels': 1
            'output_channels': 64
            'convolution_stride': 1
            'dropout': 0.2
            'pool_kernel_size': 2
            'pool_stride': 2
            'up_mode': 'upconv'
            'number_of_classes': 1
        }

    Returns:
        probability_map (torch.tensor): Output forward passed tensor through the U-net block
    """

217
218
219
    def __init__(self, parameters):
        super(BrainMapperUNet3D, self).__init__()

220
221
222
223
        original_input_channels = parameters['input_channels']
        original_output_channels = parameters['output_channels']

        self.encoderBlock1 = modules.EncoderBlock3D(parameters)
224
225
        parameters['input_channels'] = parameters['output_channels']
        parameters['output_channels'] = parameters['output_channels'] * 2
226
        self.encoderBlock2 = modules.EncoderBlock3D(parameters)
227
228
        parameters['input_channels'] = parameters['output_channels']
        parameters['output_channels'] = parameters['output_channels'] * 2
229
        self.encoderBlock3 = modules.EncoderBlock3D(parameters)
230
231
        parameters['input_channels'] = parameters['output_channels']
        parameters['output_channels'] = parameters['output_channels'] * 2
232
        self.encoderBlock4 = modules.EncoderBlock3D(parameters)
233
234
235

        parameters['input_channels'] = parameters['output_channels']
        parameters['output_channels'] = parameters['output_channels'] * 2
236
        self.bottleneck = modules.ConvolutionalBlock3D(parameters)
237
238
239

        parameters['input_channels'] = parameters['output_channels']
        parameters['output_channels'] = parameters['output_channels'] // 2
240
        self.decoderBlock1 = modules.DecoderBlock3D(parameters)
241
242
        parameters['input_channels'] = parameters['output_channels']
        parameters['output_channels'] = parameters['output_channels'] // 2
243
        self.decoderBlock2 = modules.DecoderBlock3D(parameters)
244
245
        parameters['input_channels'] = parameters['output_channels']
        parameters['output_channels'] = parameters['output_channels'] // 2
246
        self.decoderBlock3 = modules.DecoderBlock3D(parameters)
247
248
        parameters['input_channels'] = parameters['output_channels']
        parameters['output_channels'] = parameters['output_channels'] // 2
249
        self.decoderBlock4 = modules.DecoderBlock3D(parameters)
250
251

        parameters['input_channels'] = parameters['output_channels']
252
253
254
255
        self.classifier = modules.ClassifierBlock3D(parameters)

        parameters['input_channels'] = original_input_channels
        parameters['output_channels'] = original_output_channels
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

    def forward(self, X):
        """Forward pass for 3D U-net

        Function computing the forward pass through the 3D U-Net
        The input to the function is the dMRI map

        Args:
            X (torch.tensor): Input dMRI map, shape = (N x C x D x H x W) 

        Returns:
            probability_map (torch.tensor): Output forward passed tensor through the U-net block
        """

        Y_encoder_1, Y_np1, pool_indices1 = self.encoderBlock1.forward(X)
        Y_encoder_2, Y_np2, pool_indices2 = self.encoderBlock2.forward(
            Y_encoder_1)

        del Y_encoder_1

        Y_encoder_3, Y_np3, pool_indices3 = self.encoderBlock3.forward(
            Y_encoder_2)

        del Y_encoder_2

        Y_encoder_4, Y_np4, pool_indices4 = self.encoderBlock4.forward(
            Y_encoder_3)

        del Y_encoder_3

        Y_bottleNeck = self.bottleneck.forward(Y_encoder_4)

        del Y_encoder_4

        Y_decoder_1 = self.decoderBlock1.forward(
            Y_bottleNeck, Y_np4, pool_indices4)

        del Y_bottleNeck, Y_np4, pool_indices4
        
        Y_decoder_2 = self.decoderBlock2.forward(
            Y_decoder_1, Y_np3, pool_indices3)

        del Y_decoder_1, Y_np3, pool_indices3

        Y_decoder_3 = self.decoderBlock3.forward(
            Y_decoder_2, Y_np2, pool_indices2)

        del Y_decoder_2, Y_np2, pool_indices2

305
        Y_decoder_4 = self.decoderBlock4.forward(
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
            Y_decoder_3, Y_np1, pool_indices1)

        del Y_decoder_3, Y_np1, pool_indices1

        probability_map = self.classifier.forward(Y_decoder_4)

        del Y_decoder_4

        return probability_map

    def save(self, path):
        """Model Saver

        Function saving the model with all its parameters to a given path.
        The path must end with a *.model argument.

        Args:
            path (str): Path string
        """

        print("Saving Model... {}".format(path))
        torch.save(self, path)

    @property
    def test_if_cuda(self):
        """Cuda Test

        This function tests if the model parameters are allocated to a CUDA enabled GPU.

        Returns:
            bool: Flag indicating True if the tensor is stored on the GPU and Flase otherwhise
        """

        return next(self.parameters()).is_cuda

    def predict(self, X, device=0):
        """Post-training Output Prediction

        This function predicts the output of the of the U-net post-training

        Args:
            X (torch.tensor): input dMRI volume
            device (int/str): Device type used for training (int - GPU id, str- CPU)

        Returns:
            prediction (ndarray): predicted output after training

        """
        self.eval()  # PyToch module setting network to evaluation mode

        if type(X) is np.ndarray:
            X = torch.tensor(X, requires_grad=False).type(torch.FloatTensor)
        elif type(X) is torch.Tensor and not X.is_cuda:
            X = X.type(torch.FloatTensor).cuda(device, non_blocking=True)

        # .cuda() call transfers the densor from the CPU to the GPU if that is the case.
        # Non-blocking argument lets the caller bypas synchronization when necessary

        with torch.no_grad():  # Causes operations to have no gradients
            output = self.forward(X)

        _, idx = torch.max(output, 1)

        # We retrieve the tensor held by idx (.data), and map it to a cpu as an ndarray
        idx = idx.data.cpu().numpy()

        prediction = np.squeeze(idx)

        del X, output, idx

        return prediction

# DEPRECATED ARCHITECTURES!

class BrainMapperUNet3D_Simple(nn.Module):
    """Architecture class BrainMapper 3D U-net.

    This class contains the pytorch implementation of the U-net architecture underpinning the BrainMapper project.

    Args:
        parameters (dict): Contains information relevant parameters
        parameters = {
            'kernel_heigth': 5
            'kernel_width': 5
            'kernel_depth': 5
            'kernel_classification': 1
            'input_channels': 1
            'output_channels': 64
            'convolution_stride': 1
            'dropout': 0.2
            'pool_kernel_size': 2
            'pool_stride': 2
            'up_mode': 'upconv'
            'number_of_classes': 1
        }

    Returns:
        probability_map (torch.tensor): Output forward passed tensor through the U-net block
    """

406
    def __init__(self, parameters):
407
        super(BrainMapperUNet3D_Simple, self).__init__()
408
409
410
411
412
413
414
415
416
417
418

        # TODO: currently, architecture based on QuickNAT - need to adjust parameter values accordingly!

        self.encoderBlock1 = modules.EncoderBlock3D(parameters)
        parameters['input_channels'] = parameters['output_channels']
        self.encoderBlock2 = modules.EncoderBlock3D(parameters)
        self.encoderBlock3 = modules.EncoderBlock3D(parameters)
        self.encoderBlock4 = modules.EncoderBlock3D(parameters)

        self.bottleneck = modules.ConvolutionalBlock3D(parameters)

419
        parameters['input_channels'] = parameters['output_channels'] * 2
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
        self.decoderBlock1 = modules.DecoderBlock3D(parameters)
        self.decoderBlock2 = modules.DecoderBlock3D(parameters)
        self.decoderBlock3 = modules.DecoderBlock3D(parameters)
        self.decoderBlock4 = modules.DecoderBlock3D(parameters)

        parameters['input_channels'] = parameters['output_channels']
        self.classifier = modules.ClassifierBlock3D(parameters)

    def forward(self, X):
        """Forward pass for 3D U-net

        Function computing the forward pass through the 3D U-Net
        The input to the function is the dMRI map

        Args:
            X (torch.tensor): Input dMRI map, shape = (N x C x D x H x W) 

        Returns:
            probability_map (torch.tensor): Output forward passed tensor through the U-net block
        """

441
        Y_encoder_1, Y_np1, pool_indices1 = self.encoderBlock1.forward(X)
442
443
        Y_encoder_2, Y_np2, pool_indices2 = self.encoderBlock2.forward(
            Y_encoder_1)
444
445
446

        del Y_encoder_1

447
448
        Y_encoder_3, Y_np3, pool_indices3 = self.encoderBlock3.forward(
            Y_encoder_2)
449
450
451

        del Y_encoder_2

452
453
        Y_encoder_4, Y_np4, pool_indices4 = self.encoderBlock4.forward(
            Y_encoder_3)
454

455
456
        del Y_encoder_3

457
458
        Y_bottleNeck = self.bottleneck.forward(Y_encoder_4)

459
460
        del Y_encoder_4

461
462
        Y_decoder_1 = self.decoderBlock1.forward(
            Y_bottleNeck, Y_np4, pool_indices4)
463
464
465

        del Y_bottleNeck, Y_np4, pool_indices4
        
466
467
        Y_decoder_2 = self.decoderBlock2.forward(
            Y_decoder_1, Y_np3, pool_indices3)
468
469
470

        del Y_decoder_1, Y_np3, pool_indices3

471
472
        Y_decoder_3 = self.decoderBlock3.forward(
            Y_decoder_2, Y_np2, pool_indices2)
473
474
475

        del Y_decoder_2, Y_np2, pool_indices2

476
        Y_decoder_4 = self.decoderBlock4.forward(
477
            Y_decoder_3, Y_np1, pool_indices1)
478

479
480
481
482
483
484
        del Y_decoder_3, Y_np1, pool_indices1

        probability_map = self.classifier.forward(Y_decoder_4)

        del Y_decoder_4

485
        probability_map = self.classifier.forward(Y_decoder_4)
486

487
        return probability_map
488

489
    def save(self, path):
Andrei-Claudiu Roibu's avatar
Andrei-Claudiu Roibu committed
490
        """Model Saver
491

Andrei-Claudiu Roibu's avatar
Andrei-Claudiu Roibu committed
492
493
        Function saving the model with all its parameters to a given path.
        The path must end with a *.model argument.
494

Andrei-Claudiu Roibu's avatar
Andrei-Claudiu Roibu committed
495
496
497
        Args:
            path (str): Path string
        """
498

Andrei-Claudiu Roibu's avatar
Andrei-Claudiu Roibu committed
499
500
        print("Saving Model... {}".format(path))
        torch.save(self, path)
501
502
503
504
505
506
507
508
509
510
511
512

    @property
    def test_if_cuda(self):
        """Cuda Test

        This function tests if the model parameters are allocated to a CUDA enabled GPU.

        Returns:
            bool: Flag indicating True if the tensor is stored on the GPU and Flase otherwhise
        """

        return next(self.parameters()).is_cuda
513
514

    def predict(self, X, device=0):
515
        """Post-training Output Prediction
516

517
518
519
520
        This function predicts the output of the of the U-net post-training

        Args:
            X (torch.tensor): input dMRI volume
521
            device (int/str): Device type used for training (int - GPU id, str- CPU)
522
523
524
525

        Returns:
            prediction (ndarray): predicted output after training

526
        """
527
        self.eval()  # PyToch module setting network to evaluation mode
528
529
530
531
532
533
534
535
536

        if type(X) is np.ndarray:
            X = torch.tensor(X, requires_grad=False).type(torch.FloatTensor)
        elif type(X) is torch.Tensor and not X.is_cuda:
            X = X.type(torch.FloatTensor).cuda(device, non_blocking=True)

        # .cuda() call transfers the densor from the CPU to the GPU if that is the case.
        # Non-blocking argument lets the caller bypas synchronization when necessary

537
        with torch.no_grad():  # Causes operations to have no gradients
538
539
540
541
            output = self.forward(X)

        _, idx = torch.max(output, 1)

542
543
544
        # We retrieve the tensor held by idx (.data), and map it to a cpu as an ndarray
        idx = idx.data.cpu().numpy()

545
546
547
548
549
550
        prediction = np.squeeze(idx)

        del X, output, idx

        return prediction

551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
if __name__ == '__main__':

    # For debugging - To be deleted later! TODO

    parameters = {
        'kernel_heigth': 5,
        'kernel_width': 5,
        'kernel_depth': 5,
        'kernel_classification': 1,
        'input_channels': 1,
        'output_channels': 64,
        'convolution_stride': 1,
        'dropout': 0.2,
        'pool_kernel_size': 2,
        'pool_stride': 2,
        'up_mode': 'upconv',
        'number_of_classes': 1
    }

    network = BrainMapperUNet3D(parameters)