DeepSCEM/data.py at main · Cyril-Meyer/DeepSCEM · 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
import numpy as np
import tifffile
import h5py


def read_image(filename, dtype=np.float32, normalize=False, binarize=False):
    """
    read an image file into a numpy array with specific dtype and normalize option.
    """
    # read data with specific library for each file format.
    data = None
    if filename.endswith('.tif') or filename.endswith('.tiff'):
        data = tifffile.imread(filename).astype(dtype)
    elif filename.endswith('.npy'):
        raise NotImplementedError('NotImplementedError')
    else:
        raise NotImplementedError('NotImplementedError')
    # normalize value in range [0, 1].
    if normalize:
        data = (data - data.min()) / (data.max() - data.min())
    # binarize value in {0, 1}.
    if binarize:
        data = (data > 0).astype(dtype)
    # convert 2D array into 3D array.
    if len(data.shape) == 2:
        data = np.expand_dims(data, axis=0)
    # check that output is 3D array.
    if len(data.shape) != 3:
        raise NotImplementedError('NotImplementedError')

    return data


def write_image(filename, data, extension='tiff'):
    if extension in ['tif', 'tiff', 'TIF', 'TIFF']:
        tifffile.imwrite(filename + f'.{extension}', data)
    else:
        raise NotImplementedError('NotImplementedError')


def read_dataset(filename, readonly=False):
    """
    read a dataset file into a h5py object.
    """
    if readonly:
        file = h5py.File(filename, 'r')
    else:
        file = h5py.File(filename, 'a')

    if 'name' not in file.attrs:
        raise ValueError('missing name attribute')
    if 'labels' not in file.attrs:
        raise ValueError('missing labels attribute')

    return file


def create_dataset(name, labels=0, filename=None):
    if filename is None:
        filename = name + '.hdf5'
    file = h5py.File(filename, 'w')

    file.attrs['name'] = name
    file.attrs['labels'] = labels
    # file.create_dataset('name', data=name)
    file.flush()

    return file


def add_sample_to_dataset(dataset, name, image, labels=None):
    sample_group = dataset.create_group(name)
    sample_group.attrs['shape'] = image.shape
    sample_group.create_dataset('image', data=image)

    for i, label in enumerate(labels):
        sample_group.create_dataset(f'label_{i:04}', data=label)

    dataset.flush()


def crop_sample(dataset, sample_name, new_name, z_min, z_max, y_min, y_max, x_min, x_max):
    sample_group = dataset.create_group(new_name)
    sample_group.attrs['shape'] = [z_max - z_min, y_max - y_min, x_max - x_min]

    for s in dataset[sample_name].keys():
        data = np.array(dataset[sample_name][s])
        sample_group.create_dataset(f'{s}', data=data[z_min:z_max, y_min:y_max, x_min:x_max])

    dataset.flush()


def add_prediction_to_dataset(dataset, name, image, prediction):
    sample_group = dataset.create_group(name)
    sample_group.attrs['shape'] = image.shape
    sample_group.create_dataset('image', data=image)

    for i in range(prediction.shape[-1]):
        sample_group.create_dataset(f'prediction_{i:04}', data=prediction[:, :, :, i])

    dataset.flush()


def remove_sample_from_dataset(dataset, name):
    del dataset[name]
    dataset.flush()


def copy_dataset(dataset, filename):
    # inspiration: https://stackoverflow.com/a/53010788
    fd = h5py.File(filename, 'w')
    fs = dataset

    for a in fs.attrs:
        fd.attrs[a] = fs.attrs[a]
    for d in fs:
        # SFS_TRANSITION ? For now, we juste reuse working code.
        if not 'SFS_TRANSITION' in d: fs.copy(d, fd)

    fd.close()