Glimpse/data_loader.py at main · swing-research/Glimpse · GitHub

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
import torch
from utils import *
import numpy as np
import os
import imageio
import torch.nn.functional as F
from skimage.transform import iradon, radon

class CT_dataset(torch.utils.data.Dataset):

    def __init__(self, directory, network = 'glimpse'):

        self.directory = directory

        self.name_list = sorted(os.listdir(self.directory))
        self.network = network


    def __len__(self):
        return len(self.name_list)

    def __getitem__(self, idx):
        if torch.is_tensor(idx):
            idx = idx.tolist()

        file_name = self.name_list[idx]
        file = np.load(os.path.join(self.directory,file_name))
        image = file['image']
        image = torch.tensor(image, dtype = torch.float32)


        if self.network == 'unet':

            fbp = file['fbp']
            fbp = torch.tensor(fbp, dtype = torch.float32)[None,...]
            return image[None,...], fbp

        elif self.network == 'iradon':
            sinogram = file['sinogram']
            sinogram = torch.tensor(sinogram, dtype = torch.float32)

            return image[None,...], sinogram


        else:
            sinogram = file['sinogram']
            sinogram = torch.tensor(sinogram, dtype = torch.float32)

            image = image.reshape(-1, 1)
            return image, sinogram


class CT_odl(torch.utils.data.Dataset):

    def __init__(self, directory):

        self.directory = directory
        self.name_list = sorted(os.listdir(self.directory))

    def __len__(self):
        return len(self.name_list)

    def __getitem__(self, idx):
        if torch.is_tensor(idx):
            idx = idx.tolist()

        file_name = self.name_list[idx]
        data = np.load(os.path.join(self.directory,file_name))
        image = data['image']
        sinogram = data['sinogram']
        fbp = data['fbp']

        image = torch.tensor(image, dtype = torch.float32)[None,...]
        fbp = torch.tensor(fbp, dtype = torch.float32)[None,...]
        sinogram = torch.tensor(sinogram, dtype = torch.float32)[None,...]
        # print(image.shape, fbp.shape, sinogram.shape)

        ################ Remove it later
        # image = F.interpolate(image[None,...], size = image.shape[1]* (256//128),
        #                 mode = 'bilinear',
        #                 antialias= True,
        #                 align_corners= True)[0]


        # fbp = F.interpolate(fbp[None,...], size = fbp.shape[1]* (256//128),
        #                 mode = 'bilinear',
        #                 antialias= True,
        #                 align_corners= True)[0]


        # sinogram = F.interpolate(sinogram[None,...], size = (sinogram.shape[1],363),
        #                 mode = 'bilinear',
        #                 antialias= True,
        #                 align_corners= True)[0]

        # print(image.shape, fbp.shape, sinogram.shape)

        return image, sinogram, fbp


class CT_images(torch.utils.data.Dataset):

    def __init__(self, directory, image_size, theta_actual, theta_init,
                  noise_snr = 30, unet = False, subset = 'train'):

        self.directory = directory

        self.name_list = sorted(os.listdir(self.directory))
        self.image_size = image_size
        self.unet = unet
        self.theta_actual = theta_actual
        self.theta_init = theta_init
        self.noise_snr = noise_snr
        self.subset = subset


    def __len__(self):
        return len(self.name_list)

    def __getitem__(self, idx):
        if torch.is_tensor(idx):
            idx = idx.tolist()

        np.random.seed(idx + len(self.name_list))

        file_name = self.name_list[idx]
        if self.subset == 'ood':
            image = imageio.imread(os.path.join(self.directory,file_name))
            image = (image/255.0)

        elif self.subset == 'div2k':
            image = imageio.imread(os.path.join(self.directory,file_name))
            image = (image/255.0)
            # print(image.shape)
            image = np.mean(image, axis = 3)

        else:
            image = np.load(os.path.join(self.directory,file_name))


        image = torch.tensor(image, dtype = torch.float32)[None,None]
        image = F.interpolate(image, size = self.image_size,
                                mode = 'bilinear',
                                antialias= True,
                                align_corners= True)[0,0].cpu().detach().numpy()

        sinogram = radon(image, theta=self.theta_actual, circle= False)
        noise_sigma = 10**(-self.noise_snr/20.0)*np.sqrt(np.mean(np.sum(
            np.square(np.reshape(sinogram, (1 , -1))) , -1)))
        noise = np.random.normal(loc = 0,
                                 scale = noise_sigma,
                                 size = np.shape(sinogram))/np.sqrt(np.prod(np.shape(sinogram)))
        sinogram += noise

        image = torch.tensor(image, dtype = torch.float32)

        # print(image.shape, sinogram.shape)


        if self.unet:

            fbp = iradon(sinogram, theta= self.theta_init, circle= False)
            fbp = torch.tensor(fbp, dtype = torch.float32)[None,...]
            return image[None,...], fbp

        else:
            sinogram = torch.tensor(sinogram, dtype = torch.float32)
            image = image.reshape(-1, 1)
            return image, sinogram