File size: 7,684 Bytes
12d2e9e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
"""
Copyright 2021, Dana-Farber Cancer Institute and Weill Cornell Medicine
License: GNU GPL 2.0
"""

import urllib
import zipfile

import cv2
import numpy as np
import pytest

from pathml.datasets import PanNukeDataModule
from pathml.datasets.pannuke import PanNukeDataset
from pathml.datasets.utils import pannuke_multiclass_mask_to_nucleus_mask


def create_fake_pannuke_data(target_dir, n_fold=16):
    """
    create some fake images and masks in target_dir/images and target_dir/masks

    Args:
        target_dir (pathlib.Path): directory where to save the images and masks. 'images' and 'masks' subdirectories
            will be created here.
        n_fold (int): number of images and masks per fold
    """
    folds = [1, 2, 3]
    tissue_types = ["breast", "colon", "head-neck"]

    imdir = target_dir / "images"
    maskdir = target_dir / "masks"

    imdir.mkdir()
    maskdir.mkdir()

    for fold_ix in folds:
        for i in range(n_fold):
            im = np.random.randint(low=2, high=254, size=(256, 256, 3), dtype=np.uint8)
            mask = np.random.randint(low=0, high=10, size=(6, 256, 256), dtype=np.uint8)
            tissue_type = np.random.choice(tissue_types)

            im_fname = imdir / f"fold{fold_ix}_{i}_{tissue_type}.png"
            im_fname = str(im_fname.resolve())
            mask_fname = maskdir / f"fold{fold_ix}_{i}_{tissue_type}.npy"
            mask_fname = str(mask_fname.resolve())

            cv2.imwrite(im_fname, im)
            np.save(mask_fname, mask)


@pytest.mark.parametrize("fold", [1, 2, 3, None])
@pytest.mark.parametrize("nucleus_type_labels", [True, False])
def test_pannuke_dataset_sizes(tmp_path, fold, nucleus_type_labels):
    n_fold = 16
    create_fake_pannuke_data(tmp_path, n_fold=n_fold)

    pannuke_dataset = PanNukeDataset(
        data_dir=tmp_path, fold_ix=fold, nucleus_type_labels=nucleus_type_labels
    )

    # check size of dataset
    if fold in [1, 2, 3]:
        assert len(pannuke_dataset) == n_fold
    else:
        assert len(pannuke_dataset) == 3 * n_fold

    # check shapes of individual elements
    im, mask, lab = pannuke_dataset[0]
    assert im.shape == (3, 256, 256)

    if nucleus_type_labels:
        assert mask.shape == (6, 256, 256)
    else:
        assert mask.shape == (256, 256)


def create_fake_pannuke_data_raw(target_dir, fold_size=16):
    """
    Create some fake raw data that mimics file structure of what is downloaded from PanNuke website:
    https://warwick.ac.uk/fac/sci/dcs/research/tia/data/pannuke/

    Args:
        target_dir (pathlib.Path): directory where to save the data.
        fold_size (int): number of images and masks per fold
    """
    folds = [1, 2, 3]
    tissue_types = ["breast", "colon", "head-neck"]

    for fold_ix in folds:
        # create the directories
        images_dir = target_dir / f"Fold {fold_ix}" / "images" / f"fold{fold_ix}"
        masks_dir = target_dir / f"Fold {fold_ix}" / "masks" / f"fold{fold_ix}"
        images_dir.mkdir(parents=True)
        masks_dir.mkdir(parents=True)

        # create the fake data
        types_fold = np.random.choice(tissue_types, size=fold_size)
        masks = np.random.randint(low=0, high=10, size=(fold_size, 256, 256, 6))
        ims = np.random.randint(low=0, high=254, size=(fold_size, 256, 256, 3))

        # write the data
        np.save(file=str(images_dir / "images.npy"), arr=ims)
        np.save(file=str(images_dir / "types.npy"), arr=types_fold)
        np.save(file=str(masks_dir / "masks.npy"), arr=masks)


def test_process_downloaded_pannuke(tmp_path):
    """Test the post-processing of the pannuke raw data"""
    # make fake data
    fold_size = 16
    create_fake_pannuke_data_raw(tmp_path, fold_size=fold_size)

    # process the fake data
    PanNukeDataModule._process_downloaded_pannuke(tmp_path)

    # check everything
    imdir = tmp_path / "images"
    maskdir = tmp_path / "masks"

    assert imdir.is_dir()
    assert maskdir.is_dir()

    assert len(list(imdir.glob("*"))) == 3 * fold_size
    assert len(list(maskdir.glob("*"))) == 3 * fold_size

    for fold_ix in [1, 2, 3]:
        assert len(list(imdir.glob(f"fold{fold_ix}*"))) == fold_size
        assert len(list(maskdir.glob(f"fold{fold_ix}*"))) == fold_size


@pytest.mark.parametrize("raw_data", [True, False])
@pytest.mark.parametrize("hovernet_preprocess", [True, False])
@pytest.mark.parametrize("split", [1, 2, 3, None])
@pytest.mark.parametrize("nucleus_type_labels", [True, False])
def test_pannuke_datamodule(
    tmp_path, split, nucleus_type_labels, hovernet_preprocess, raw_data
):
    # make fake data
    # if raw_data, then also test processing step
    fold_size = 8

    if raw_data:
        create_fake_pannuke_data_raw(tmp_path, fold_size=fold_size)
        # process the fake data
        PanNukeDataModule._process_downloaded_pannuke(tmp_path)
    else:
        create_fake_pannuke_data(tmp_path, n_fold=fold_size)

    batch_size = 4
    pannuke = PanNukeDataModule(
        data_dir=tmp_path,
        nucleus_type_labels=nucleus_type_labels,
        split=split,
        download=False,
        transforms=None,
        batch_size=batch_size,
        hovernet_preprocess=hovernet_preprocess,
    )

    train = pannuke.train_dataloader
    valid = pannuke.valid_dataloader
    test = pannuke.test_dataloader

    for loader in [train, test, valid]:
        # make sure everything is correct dimensions
        if hovernet_preprocess:
            im, mask, hv, tissue_types = next(iter(loader))
            assert hv.shape == (batch_size, 2, 256, 256)
        else:
            im, mask, tissue_types = next(iter(loader))

        assert im.shape == (batch_size, 3, 256, 256)
        if nucleus_type_labels:
            assert mask.shape == (batch_size, 6, 256, 256)
        else:
            assert mask.shape == (batch_size, 256, 256)

        assert len(tissue_types) == batch_size and all(
            [isinstance(t, str) for t in tissue_types]
        )


def test_clean_up_download_pannuke(tmp_path):
    # first create the files and dirs to delete
    for fold_ix in [1, 2, 3]:
        with zipfile.ZipFile(tmp_path / f"fold_{fold_ix}.zip", "w") as myzip:
            myzip.writestr("fake_pannuke_data.txt", "NYE 2020 - happy new year!")
        downloaded_dir = tmp_path / f"Fold {fold_ix}"
        downloaded_dir.mkdir()
        # add some data inside the dirs (test for issue #53)
        data = np.random.randint(low=0, high=10, size=(8, 256, 256, 6))
        np.save(file=str(downloaded_dir / "data.npy"), arr=data)

    # now call cleanup
    PanNukeDataModule._clean_up_download_pannuke(tmp_path)

    # now make sure that the files/dirs were deleted
    for fold_ix in [1, 2, 3]:
        zfile = tmp_path / f"fold_{fold_ix}.zip"
        downloaded_dir = tmp_path / f"Fold {fold_ix}"
        assert not zfile.exists()
        assert not downloaded_dir.exists()


def check_pannuke_data_urls():
    # make sure that the urls for the pannuke data are still valid!
    for fold_ix in [1, 2, 3]:
        url = f"https://warwick.ac.uk/fac/sci/dcs/research/tia/data/pannuke/fold_{fold_ix}.zip"
        r = urllib.request.urlopen(url)
        # HTTP status code 200 means "OK"
        assert r.getcode() == 200


def check_wrong_path_download_false_fails():
    with pytest.raises(AssertionError):
        PanNukeDataModule(data_dir="wrong/path/to/pannuke", download=False)


def test_pannuke_multiclass_mask_to_nucleus_mask():
    mask = np.random.randint(low=0, high=10, size=(6, 256, 256), dtype=np.uint8)
    mask_1c = pannuke_multiclass_mask_to_nucleus_mask(mask)
    assert mask_1c.shape == (256, 256)


# TODO add tests for _download_pannuke()