[tbx11k] Implements database, improves documentation

src/ptbench/data/tbx11k/v2_fold_3.py

+# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
+#
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+from .datamodule import DataModule
+
+datamodule = DataModule("v2-fold-3.json")

src/ptbench/data/tbx11k/v2_fold_4.py

+# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
+#
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+from .datamodule import DataModule
+
+datamodule = DataModule("v2-fold-4.json")

src/ptbench/data/tbx11k/v2_fold_5.py

+# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
+#
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+from .datamodule import DataModule
+
+datamodule = DataModule("v2-fold-5.json")

src/ptbench/data/tbx11k/v2_fold_6.py

+# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
+#
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+from .datamodule import DataModule
+
+datamodule = DataModule("v2-fold-6.json")

src/ptbench/data/tbx11k/v2_fold_7.py

+# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
+#
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+from .datamodule import DataModule
+
+datamodule = DataModule("v2-fold-7.json")

src/ptbench/data/tbx11k/v2_fold_8.py

+# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
+#
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+from .datamodule import DataModule
+
+datamodule = DataModule("v2-fold-8.json")

src/ptbench/data/tbx11k/v2_fold_9.py

+# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
+#
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+from .datamodule import DataModule
+
+datamodule = DataModule("v2-fold-9.json")

src/ptbench/data/tbx11k/v2_others_vs_atb.py

+# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
+#
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+from .datamodule import DataModule
+
+datamodule = DataModule("v2-others-vs-atb.json")

src/ptbench/data/tbx11k_simplified/__init__.py

-# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
-#
-# SPDX-License-Identifier: GPL-3.0-or-later
-
-"""TBX11K simplified dataset for computer-aided diagnosis.
-
-The TBX11K database has been established to foster research
-in computer-aided diagnosis of pulmonary diseases with a special
-focus on tuberculosis (aTB). The dataset was specifically
-designed to be used with CNNs. It contains 11,000 chest X-ray
-images, each of a unique patient. They were labeled by expert
-radiologists with 5 - 10+ years of experience. Possible labels
-are: "healthy", "active TB", "latent TB", and "sick & non-tb".
-The version of the dataset used in this benchmark is a simplified.
-
-* Reference: [TBX11K-SIMPLIFIED-2020]_
-* Original (released) resolution (height x width or width x height): 512 x 512
-* Split reference: none
-* Protocol ``default``:
-
-  * Training samples: 62.5% of TB and healthy CXR (including labels)
-  * Validation samples: 15.9% of TB and healthy CXR (including labels)
-  * Test samples: 21.6% of TB and healthy CXR (including labels)
-"""
-
-import importlib.resources
-import os
-
-from ...utils.rc import load_rc
-from ..dataset import JSONDataset
-from ..loader import load_pil_grayscale, make_delayed, make_delayed_bbox
-
-_protocols = [
-    importlib.resources.files(__name__).joinpath("default.json.bz2"),
-    importlib.resources.files(__name__).joinpath("fold_0.json.bz2"),
-    importlib.resources.files(__name__).joinpath("fold_1.json.bz2"),
-    importlib.resources.files(__name__).joinpath("fold_2.json.bz2"),
-    importlib.resources.files(__name__).joinpath("fold_3.json.bz2"),
-    importlib.resources.files(__name__).joinpath("fold_4.json.bz2"),
-    importlib.resources.files(__name__).joinpath("fold_5.json.bz2"),
-    importlib.resources.files(__name__).joinpath("fold_6.json.bz2"),
-    importlib.resources.files(__name__).joinpath("fold_7.json.bz2"),
-    importlib.resources.files(__name__).joinpath("fold_8.json.bz2"),
-    importlib.resources.files(__name__).joinpath("fold_9.json.bz2"),
-]
-
-_datadir = load_rc().get(
-    "datadir.tbx11k_simplified", os.path.realpath(os.curdir)
-)
-
-
-def _raw_data_loader(sample):
-    return dict(
-        data=load_pil_grayscale(os.path.join(_datadir, sample["data"])),  # type: ignore
-        label=sample["label"],
-    )
-
-
-def _raw_data_loader_bbox(sample):
-    return dict(
-        data=load_pil_grayscale(os.path.join(_datadir, sample["data"])),  # type: ignore
-        label=sample["label"],
-        bboxes=sample["bboxes"],
-    )
-
-
-def _loader(context, sample):
-    # "context" is ignored in this case - database is homogeneous
-    # we return delayed samples to avoid loading all images at once
-    return make_delayed(sample, _raw_data_loader)
-
-
-def _loader_bbox(context, sample):
-    # "context" is ignored in this case - database is homogeneous
-    # we return delayed samples to avoid loading all images at once
-    return make_delayed_bbox(sample, _raw_data_loader_bbox)
-
-
-json_dataset = JSONDataset(
-    protocols=_protocols,
-    fieldnames=("data", "label"),
-    loader=_loader,
-)
-
-json_dataset_with_bboxes = JSONDataset(
-    protocols=_protocols,
-    fieldnames=("data", "label", "bboxes"),
-    loader=_loader_bbox,
-)
-"""TBX11K simplified dataset object."""
-
-
-def _maker(protocol, RGB=False):
-    from torchvision import transforms
-
-    from .. import make_dataset
-    from ..augmentations import ElasticDeformation
-
-    post_transforms = []
-    if RGB:
-        post_transforms = [
-            transforms.Lambda(lambda x: x.convert("RGB")),
-            transforms.ToTensor(),
-        ]
-
-    return make_dataset(
-        [json_dataset.subsets(protocol)],
-        [],
-        [ElasticDeformation(p=0.8)],
-        post_transforms,
-    )

src/ptbench/data/tbx11k_simplified/default.py

-# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
-#
-# SPDX-License-Identifier: GPL-3.0-or-later
-
-"""TBX11k simplified dataset for TB detection (default protocol)
-
-* Split reference: first 62.5% of TB and healthy CXR for "train" 15.9% for
-* "validation", 21.6% for "test"
-* This split only consists of healthy and active TB samples
-* "Latent TB" or "sick & non-TB" samples are not included in this configuration
-* This configuration resolution: 512 x 512 (default)
-* See :py:mod:`ptbench.data.tbx11k_simplified` for dataset details
-"""
-
-from clapper.logging import setup
-
-from .. import return_subsets
-from ..base_datamodule import BaseDataModule
-from . import _maker
-
-logger = setup(__name__.split(".")[0], format="%(levelname)s: %(message)s")
-
-
-class DefaultModule(BaseDataModule):
-    def __init__(
-        self,
-        train_batch_size=1,
-        predict_batch_size=1,
-        drop_incomplete_batch=False,
-        multiproc_kwargs=None,
-    ):
-        super().__init__(
-            train_batch_size=train_batch_size,
-            predict_batch_size=predict_batch_size,
-            drop_incomplete_batch=drop_incomplete_batch,
-            multiproc_kwargs=multiproc_kwargs,
-        )
-
-    def setup(self, stage: str):
-        self.dataset = _maker("default")
-        (
-            self.train_dataset,
-            self.validation_dataset,
-            self.extra_validation_datasets,
-            self.predict_dataset,
-        ) = return_subsets(self.dataset)
-
-
-datamodule = DefaultModule

src/ptbench/data/tbx11k_simplified/fold_0.py

-# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
-#
-# SPDX-License-Identifier: GPL-3.0-or-later
-
-"""TBX11k simplified dataset for TB detection (default protocol)
-
-* Split reference: first 62.5% of TB and healthy CXR for "train" 15.9% for
-* "validation", 21.6% for "test"
-* This split only consists of healthy and active TB samples
-* "Latent TB" or "sick & non-TB" samples are not included in this configuration
-* This configuration resolution: 512 x 512 (default)
-* See :py:mod:`ptbench.data.tbx11k_simplified` for dataset details
-"""
-
-from clapper.logging import setup
-
-from .. import return_subsets
-from ..base_datamodule import BaseDataModule
-from . import _maker
-
-logger = setup(__name__.split(".")[0], format="%(levelname)s: %(message)s")
-
-
-class Fold0Module(BaseDataModule):
-    def __init__(
-        self,
-        train_batch_size=1,
-        predict_batch_size=1,
-        drop_incomplete_batch=False,
-        multiproc_kwargs=None,
-    ):
-        super().__init__(
-            train_batch_size=train_batch_size,
-            predict_batch_size=predict_batch_size,
-            drop_incomplete_batch=drop_incomplete_batch,
-            multiproc_kwargs=multiproc_kwargs,
-        )
-
-    def setup(self, stage: str):
-        self.dataset = _maker("fold_0")
-        (
-            self.train_dataset,
-            self.validation_dataset,
-            self.extra_validation_datasets,
-            self.predict_dataset,
-        ) = return_subsets(self.dataset)
-
-
-datamodule = Fold0Module

src/ptbench/data/tbx11k_simplified/fold_1.py

-# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
-#
-# SPDX-License-Identifier: GPL-3.0-or-later
-
-"""TBX11k simplified dataset for TB detection (default protocol)
-
-* Split reference: first 62.5% of TB and healthy CXR for "train" 15.9% for
-* "validation", 21.6% for "test"
-* This split only consists of healthy and active TB samples
-* "Latent TB" or "sick & non-TB" samples are not included in this configuration
-* This configuration resolution: 512 x 512 (default)
-* See :py:mod:`ptbench.data.tbx11k_simplified` for dataset details
-"""
-
-from clapper.logging import setup
-
-from .. import return_subsets
-from ..base_datamodule import BaseDataModule
-from . import _maker
-
-logger = setup(__name__.split(".")[0], format="%(levelname)s: %(message)s")
-
-
-class Fold0Module(BaseDataModule):
-    def __init__(
-        self,
-        train_batch_size=1,
-        predict_batch_size=1,
-        drop_incomplete_batch=False,
-        multiproc_kwargs=None,
-    ):
-        super().__init__(
-            train_batch_size=train_batch_size,
-            predict_batch_size=predict_batch_size,
-            drop_incomplete_batch=drop_incomplete_batch,
-            multiproc_kwargs=multiproc_kwargs,
-        )
-
-    def setup(self, stage: str):
-        self.dataset = _maker("fold_1")
-        (
-            self.train_dataset,
-            self.validation_dataset,
-            self.extra_validation_datasets,
-            self.predict_dataset,
-        ) = return_subsets(self.dataset)
-
-
-datamodule = Fold0Module

src/ptbench/data/tbx11k_simplified/fold_2.py

-# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
-#
-# SPDX-License-Identifier: GPL-3.0-or-later
-
-"""TBX11k simplified dataset for TB detection (default protocol)
-
-* Split reference: first 62.5% of TB and healthy CXR for "train" 15.9% for
-* "validation", 21.6% for "test"
-* This split only consists of healthy and active TB samples
-* "Latent TB" or "sick & non-TB" samples are not included in this configuration
-* This configuration resolution: 512 x 512 (default)
-* See :py:mod:`ptbench.data.tbx11k_simplified` for dataset details
-"""
-
-from clapper.logging import setup
-
-from .. import return_subsets
-from ..base_datamodule import BaseDataModule
-from . import _maker
-
-logger = setup(__name__.split(".")[0], format="%(levelname)s: %(message)s")
-
-
-class Fold0Module(BaseDataModule):
-    def __init__(
-        self,
-        train_batch_size=1,
-        predict_batch_size=1,
-        drop_incomplete_batch=False,
-        multiproc_kwargs=None,
-    ):
-        super().__init__(
-            train_batch_size=train_batch_size,
-            predict_batch_size=predict_batch_size,
-            drop_incomplete_batch=drop_incomplete_batch,
-            multiproc_kwargs=multiproc_kwargs,
-        )
-
-    def setup(self, stage: str):
-        self.dataset = _maker("fold_2")
-        (
-            self.train_dataset,
-            self.validation_dataset,
-            self.extra_validation_datasets,
-            self.predict_dataset,
-        ) = return_subsets(self.dataset)
-
-
-datamodule = Fold0Module

src/ptbench/data/tbx11k_simplified/fold_3.py

-# SPDX-FileCopyrightText: Copyright © 2023 Idiap Research Institute <contact@idiap.ch>
-#
-# SPDX-License-Identifier: GPL-3.0-or-later
-
-"""TBX11k simplified dataset for TB detection (default protocol)
-
-* Split reference: first 62.5% of TB and healthy CXR for "train" 15.9% for
-* "validation", 21.6% for "test"
-* This split only consists of healthy and active TB samples
-* "Latent TB" or "sick & non-TB" samples are not included in this configuration
-* This configuration resolution: 512 x 512 (default)
-* See :py:mod:`ptbench.data.tbx11k_simplified` for dataset details
-"""
-
-from clapper.logging import setup
-
-from .. import return_subsets
-from ..base_datamodule import BaseDataModule
-from . import _maker
-
-logger = setup(__name__.split(".")[0], format="%(levelname)s: %(message)s")
-
-
-class Fold0Module(BaseDataModule):
-    def __init__(
-        self,
-        train_batch_size=1,
-        predict_batch_size=1,
-        drop_incomplete_batch=False,
-        multiproc_kwargs=None,
-    ):
-        super().__init__(
-            train_batch_size=train_batch_size,
-            predict_batch_size=predict_batch_size,
-            drop_incomplete_batch=drop_incomplete_batch,
-            multiproc_kwargs=multiproc_kwargs,
-        )
-
-    def setup(self, stage: str):
-        self.dataset = _maker("fold_3")
-        (
-            self.train_dataset,
-            self.validation_dataset,
-            self.extra_validation_datasets,
-            self.predict_dataset,
-        ) = return_subsets(self.dataset)
-
-
-datamodule = Fold0Module