File size: 10,596 Bytes
e4b9a7b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# Copyright 2020 MONAI Consortium
# 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 typing import Dict, Optional, Sequence, Union

import torch
import torch.nn as nn
import torch.nn.functional as F

from monai.networks.layers.convutils import same_padding
from monai.networks.layers.factories import Conv, Dropout, Norm
from monai.utils import Activation, ChannelMatching, Normalisation

SUPPORTED_NORM = {
    Normalisation.BATCH: lambda spatial_dims: Norm[Norm.BATCH, spatial_dims],
    Normalisation.INSTANCE: lambda spatial_dims: Norm[Norm.INSTANCE, spatial_dims],
}
SUPPORTED_ACTI = {Activation.RELU: nn.ReLU, Activation.PRELU: nn.PReLU, Activation.RELU6: nn.ReLU6}
DEFAULT_LAYER_PARAMS_3D = (
    # initial conv layer
    {"name": "conv_0", "n_features": 16, "kernel_size": 3},
    # residual blocks
    {"name": "res_1", "n_features": 16, "kernels": (3, 3), "repeat": 3},
    {"name": "res_2", "n_features": 32, "kernels": (3, 3), "repeat": 3},
    {"name": "res_3", "n_features": 64, "kernels": (3, 3), "repeat": 3},
    # final conv layers
    {"name": "conv_1", "n_features": 80, "kernel_size": 1},
    {"name": "conv_2", "kernel_size": 1},
)


class ConvNormActi(nn.Module):
    def __init__(
        self,
        spatial_dims: int,
        in_channels: int,
        out_channels: int,
        kernel_size: int,
        norm_type: Optional[Union[Normalisation, str]] = None,
        acti_type: Optional[Union[Activation, str]] = None,
        dropout_prob: Optional[float] = None,
    ) -> None:
        """
        Args:
            spatial_dims: number of spatial dimensions of the input image.
            in_channels: number of input channels.
            out_channels: number of output channels.
            kernel_size: size of the convolving kernel.
            norm_type: {``"batch"``, ``"instance"``}
                Feature normalisation with batchnorm or instancenorm. Defaults to ``"batch"``.
            acti_type: {``"relu"``, ``"prelu"``, ``"relu6"``}
                Non-linear activation using ReLU or PReLU. Defaults to ``"relu"``.
            dropout_prob: probability of the feature map to be zeroed
                (only applies to the penultimate conv layer).
        """

        super(ConvNormActi, self).__init__()

        layers = nn.ModuleList()

        conv_type = Conv[Conv.CONV, spatial_dims]
        padding_size = same_padding(kernel_size)
        conv = conv_type(in_channels, out_channels, kernel_size, padding=padding_size)
        layers.append(conv)

        if norm_type is not None:
            norm_type = Normalisation(norm_type)
            layers.append(SUPPORTED_NORM[norm_type](spatial_dims)(out_channels))
        if acti_type is not None:
            acti_type = Activation(acti_type)
            layers.append(SUPPORTED_ACTI[acti_type](inplace=True))
        if dropout_prob is not None:
            dropout_type = Dropout[Dropout.DROPOUT, spatial_dims]
            layers.append(dropout_type(p=dropout_prob))
        self.layers = nn.Sequential(*layers)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return torch.as_tensor(self.layers(x))


class HighResBlock(nn.Module):
    def __init__(
        self,
        spatial_dims: int,
        in_channels: int,
        out_channels: int,
        kernels: Sequence[int] = (3, 3),
        dilation: Union[Sequence[int], int] = 1,
        norm_type: Union[Normalisation, str] = Normalisation.INSTANCE,
        acti_type: Union[Activation, str] = Activation.RELU,
        channel_matching: Union[ChannelMatching, str] = ChannelMatching.PAD,
    ) -> None:
        """
        Args:
            spatial_dims: number of spatial dimensions of the input image.
            in_channels: number of input channels.
            out_channels: number of output channels.
            kernels: each integer k in `kernels` corresponds to a convolution layer with kernel size k.
            dilation: spacing between kernel elements.
            norm_type: {``"batch"``, ``"instance"``}
                Feature normalisation with batchnorm or instancenorm. Defaults to ``"instance"``.
            acti_type: {``"relu"``, ``"prelu"``, ``"relu6"``}
                Non-linear activation using ReLU or PReLU. Defaults to ``"relu"``.
            channel_matching: {``"pad"``, ``"project"``}
                Specifies handling residual branch and conv branch channel mismatches. Defaults to ``"pad"``.

                - ``"pad"``: with zero padding.
                - ``"project"``: with a trainable conv with kernel size.

        Raises:
            ValueError: When ``channel_matching=pad`` and ``in_channels > out_channels``. Incompatible values.

        """
        super(HighResBlock, self).__init__()
        conv_type = Conv[Conv.CONV, spatial_dims]
        norm_type = Normalisation(norm_type)
        acti_type = Activation(acti_type)

        self.project, self.pad = None, None
        if in_channels != out_channels:
            channel_matching = ChannelMatching(channel_matching)
            if channel_matching == ChannelMatching.PROJECT:
                self.project = conv_type(in_channels, out_channels, kernel_size=1)
            if channel_matching == ChannelMatching.PAD:
                if in_channels > out_channels:
                    raise ValueError('Incompatible values: channel_matching="pad" and in_channels > out_channels.')
                pad_1 = (out_channels - in_channels) // 2
                pad_2 = out_channels - in_channels - pad_1
                pad = [0, 0] * spatial_dims + [pad_1, pad_2] + [0, 0]
                self.pad = lambda input: F.pad(input, pad)

        layers = nn.ModuleList()
        _in_chns, _out_chns = in_channels, out_channels
        for kernel_size in kernels:
            layers.append(SUPPORTED_NORM[norm_type](spatial_dims)(_in_chns))
            layers.append(SUPPORTED_ACTI[acti_type](inplace=True))
            layers.append(
                conv_type(
                    _in_chns, _out_chns, kernel_size, padding=same_padding(kernel_size, dilation), dilation=dilation
                )
            )
            _in_chns = _out_chns
        self.layers = nn.Sequential(*layers)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        x_conv: torch.Tensor = self.layers(x)
        if self.project is not None:
            return x_conv + torch.as_tensor(self.project(x))
        if self.pad is not None:
            return x_conv + torch.as_tensor(self.pad(x))
        return x_conv + x


class HighResNet(nn.Module):
    """
    Reimplementation of highres3dnet based on
    Li et al., "On the compactness, efficiency, and representation of 3D
    convolutional networks: Brain parcellation as a pretext task", IPMI '17

    Adapted from:
    https://github.com/NifTK/NiftyNet/blob/v0.6.0/niftynet/network/highres3dnet.py
    https://github.com/fepegar/highresnet

    Args:
        spatial_dims: number of spatial dimensions of the input image.
        in_channels: number of input channels.
        out_channels: number of output channels.
        norm_type: {``"batch"``, ``"instance"``}
            Feature normalisation with batchnorm or instancenorm. Defaults to ``"batch"``.
        acti_type: {``"relu"``, ``"prelu"``, ``"relu6"``}
            Non-linear activation using ReLU or PReLU. Defaults to ``"relu"``.
        dropout_prob: probability of the feature map to be zeroed
            (only applies to the penultimate conv layer).
        layer_params: specifying key parameters of each layer/block.
    """

    def __init__(
        self,
        spatial_dims: int = 3,
        in_channels: int = 1,
        out_channels: int = 1,
        norm_type: Union[Normalisation, str] = Normalisation.BATCH,
        acti_type: Union[Activation, str] = Activation.RELU,
        dropout_prob: Optional[float] = None,
        layer_params: Sequence[Dict] = DEFAULT_LAYER_PARAMS_3D,
    ) -> None:

        super(HighResNet, self).__init__()
        blocks = nn.ModuleList()

        # intial conv layer
        params = layer_params[0]
        _in_chns, _out_chns = in_channels, params["n_features"]
        blocks.append(
            ConvNormActi(
                spatial_dims,
                _in_chns,
                _out_chns,
                kernel_size=params["kernel_size"],
                norm_type=norm_type,
                acti_type=acti_type,
                dropout_prob=None,
            )
        )

        # residual blocks
        for (idx, params) in enumerate(layer_params[1:-2]):  # res blocks except the 1st and last two conv layers.
            _in_chns, _out_chns = _out_chns, params["n_features"]
            _dilation = 2 ** idx
            for _ in range(params["repeat"]):
                blocks.append(
                    HighResBlock(
                        spatial_dims,
                        _in_chns,
                        _out_chns,
                        params["kernels"],
                        dilation=_dilation,
                        norm_type=norm_type,
                        acti_type=acti_type,
                    )
                )
                _in_chns = _out_chns

        # final conv layers
        params = layer_params[-2]
        _in_chns, _out_chns = _out_chns, params["n_features"]
        blocks.append(
            ConvNormActi(
                spatial_dims,
                _in_chns,
                _out_chns,
                kernel_size=params["kernel_size"],
                norm_type=norm_type,
                acti_type=acti_type,
                dropout_prob=dropout_prob,
            )
        )

        params = layer_params[-1]
        _in_chns = _out_chns
        blocks.append(
            ConvNormActi(
                spatial_dims,
                _in_chns,
                out_channels,
                kernel_size=params["kernel_size"],
                norm_type=norm_type,
                acti_type=None,
                dropout_prob=None,
            )
        )

        self.blocks = nn.Sequential(*blocks)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return torch.as_tensor(self.blocks(x))