datasets.py

# Copyright (c) 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 __future__ import annotations

import os
import shutil
import sys
import warnings
from collections.abc import Callable, Sequence
from pathlib import Path
from typing import Any

import numpy as np

from monai.apps.tcia import (
    DCM_FILENAME_REGEX,
    download_tcia_series_instance,
    get_tcia_metadata,
    get_tcia_ref_uid,
    match_tcia_ref_uid_in_study,
)
from monai.apps.utils import download_and_extract
from monai.config.type_definitions import PathLike
from monai.data import (
    CacheDataset,
    PydicomReader,
    load_decathlon_datalist,
    load_decathlon_properties,
    partition_dataset,
    select_cross_validation_folds,
)
from monai.transforms import LoadImaged, Randomizable
from monai.utils import ensure_tuple

__all__ = ["MedNISTDataset", "DecathlonDataset", "CrossValidation", "TciaDataset"]


class MedNISTDataset(Randomizable, CacheDataset):
    """
    The Dataset to automatically download MedNIST data and generate items for training, validation or test.
    It's based on `CacheDataset` to accelerate the training process.

    Args:
        root_dir: target directory to download and load MedNIST dataset.
        section: expected data section, can be: `training`, `validation` or `test`.
        transform: transforms to execute operations on input data.
        download: whether to download and extract the MedNIST from resource link, default is False.
            if expected file already exists, skip downloading even set it to True.
            user can manually copy `MedNIST.tar.gz` file or `MedNIST` folder to root directory.
        seed: random seed to randomly split training, validation and test datasets, default is 0.
        val_frac: percentage of validation fraction in the whole dataset, default is 0.1.
        test_frac: percentage of test fraction in the whole dataset, default is 0.1.
        cache_num: number of items to be cached. Default is `sys.maxsize`.
            will take the minimum of (cache_num, data_length x cache_rate, data_length).
        cache_rate: percentage of cached data in total, default is 1.0 (cache all).
            will take the minimum of (cache_num, data_length x cache_rate, data_length).
        num_workers: the number of worker threads if computing cache in the initialization.
            If num_workers is None then the number returned by os.cpu_count() is used.
            If a value less than 1 is specified, 1 will be used instead.
        progress: whether to display a progress bar when downloading dataset and computing the transform cache content.
        copy_cache: whether to `deepcopy` the cache content before applying the random transforms,
            default to `True`. if the random transforms don't modify the cached content
            (for example, randomly crop from the cached image and deepcopy the crop region)
            or if every cache item is only used once in a `multi-processing` environment,
            may set `copy=False` for better performance.
        as_contiguous: whether to convert the cached NumPy array or PyTorch tensor to be contiguous.
            it may help improve the performance of following logic.
        runtime_cache: whether to compute cache at the runtime, default to `False` to prepare
            the cache content at initialization. See: :py:class:`monai.data.CacheDataset`.

    Raises:
        ValueError: When ``root_dir`` is not a directory.
        RuntimeError: When ``dataset_dir`` doesn't exist and downloading is not selected (``download=False``).

    """

    resource = "https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/MedNIST.tar.gz"
    md5 = "0bc7306e7427e00ad1c5526a6677552d"
    compressed_file_name = "MedNIST.tar.gz"
    dataset_folder_name = "MedNIST"

    def __init__(
        self,
        root_dir: PathLike,
        section: str,
        transform: Sequence[Callable] | Callable = (),
        download: bool = False,
        seed: int = 0,
        val_frac: float = 0.1,
        test_frac: float = 0.1,
        cache_num: int = sys.maxsize,
        cache_rate: float = 1.0,
        num_workers: int | None = 1,
        progress: bool = True,
        copy_cache: bool = True,
        as_contiguous: bool = True,
        runtime_cache: bool = False,
    ) -> None:
        root_dir = Path(root_dir)
        if not root_dir.is_dir():
            raise ValueError("Root directory root_dir must be a directory.")
        self.section = section
        self.val_frac = val_frac
        self.test_frac = test_frac
        self.set_random_state(seed=seed)
        tarfile_name = root_dir / self.compressed_file_name
        dataset_dir = root_dir / self.dataset_folder_name
        self.num_class = 0
        if download:
            download_and_extract(
                url=self.resource,
                filepath=tarfile_name,
                output_dir=root_dir,
                hash_val=self.md5,
                hash_type="md5",
                progress=progress,
            )

        if not dataset_dir.is_dir():
            raise RuntimeError(
                f"Cannot find dataset directory: {dataset_dir}, please use download=True to download it."
            )
        data = self._generate_data_list(dataset_dir)
        if transform == ():
            transform = LoadImaged("image")
        CacheDataset.__init__(
            self,
            data=data,
            transform=transform,
            cache_num=cache_num,
            cache_rate=cache_rate,
            num_workers=num_workers,
            progress=progress,
            copy_cache=copy_cache,
            as_contiguous=as_contiguous,
            runtime_cache=runtime_cache,
        )

    def randomize(self, data: np.ndarray) -> None:
        self.R.shuffle(data)

    def get_num_classes(self) -> int:
        """Get number of classes."""
        return self.num_class

    def _generate_data_list(self, dataset_dir: PathLike) -> list[dict]:
        """
        Raises:
            ValueError: When ``section`` is not one of ["training", "validation", "test"].

        """
        dataset_dir = Path(dataset_dir)
        class_names = sorted(f"{x.name}" for x in dataset_dir.iterdir() if x.is_dir())  # folder name as the class name
        self.num_class = len(class_names)
        image_files = [[f"{x}" for x in (dataset_dir / class_names[i]).iterdir()] for i in range(self.num_class)]
        num_each = [len(image_files[i]) for i in range(self.num_class)]
        image_files_list = []
        image_class = []
        class_name = []
        for i in range(self.num_class):
            image_files_list.extend(image_files[i])
            image_class.extend([i] * num_each[i])
            class_name.extend([class_names[i]] * num_each[i])

        length = len(image_files_list)
        indices = np.arange(length)
        self.randomize(indices)

        test_length = int(length * self.test_frac)
        val_length = int(length * self.val_frac)
        if self.section == "test":
            section_indices = indices[:test_length]
        elif self.section == "validation":
            section_indices = indices[test_length : test_length + val_length]
        elif self.section == "training":
            section_indices = indices[test_length + val_length :]
        else:
            raise ValueError(
                f'Unsupported section: {self.section}, available options are ["training", "validation", "test"].'
            )
        # the types of label and class name should be compatible with the pytorch dataloader
        return [
            {"image": image_files_list[i], "label": image_class[i], "class_name": class_name[i]}
            for i in section_indices
        ]


class DecathlonDataset(Randomizable, CacheDataset):
    """
    The Dataset to automatically download the data of Medical Segmentation Decathlon challenge
    (http://medicaldecathlon.com/) and generate items for training, validation or test.
    It will also load these properties from the JSON config file of dataset. user can call `get_properties()`
    to get specified properties or all the properties loaded.
    It's based on :py:class:`monai.data.CacheDataset` to accelerate the training process.

    Args:
        root_dir: user's local directory for caching and loading the MSD datasets.
        task: which task to download and execute: one of list ("Task01_BrainTumour", "Task02_Heart",
            "Task03_Liver", "Task04_Hippocampus", "Task05_Prostate", "Task06_Lung", "Task07_Pancreas",
            "Task08_HepaticVessel", "Task09_Spleen", "Task10_Colon").
        section: expected data section, can be: `training`, `validation` or `test`.
        transform: transforms to execute operations on input data.
            for further usage, use `EnsureChannelFirstd` to convert the shape to [C, H, W, D].
        download: whether to download and extract the Decathlon from resource link, default is False.
            if expected file already exists, skip downloading even set it to True.
            user can manually copy tar file or dataset folder to the root directory.
        val_frac: percentage of validation fraction in the whole dataset, default is 0.2.
        seed: random seed to randomly shuffle the datalist before splitting into training and validation, default is 0.
            note to set same seed for `training` and `validation` sections.
        cache_num: number of items to be cached. Default is `sys.maxsize`.
            will take the minimum of (cache_num, data_length x cache_rate, data_length).
        cache_rate: percentage of cached data in total, default is 1.0 (cache all).
            will take the minimum of (cache_num, data_length x cache_rate, data_length).
        num_workers: the number of worker threads if computing cache in the initialization.
            If num_workers is None then the number returned by os.cpu_count() is used.
            If a value less than 1 is specified, 1 will be used instead.
        progress: whether to display a progress bar when downloading dataset and computing the transform cache content.
        copy_cache: whether to `deepcopy` the cache content before applying the random transforms,
            default to `True`. if the random transforms don't modify the cached content
            (for example, randomly crop from the cached image and deepcopy the crop region)
            or if every cache item is only used once in a `multi-processing` environment,
            may set `copy=False` for better performance.
        as_contiguous: whether to convert the cached NumPy array or PyTorch tensor to be contiguous.
            it may help improve the performance of following logic.
        runtime_cache: whether to compute cache at the runtime, default to `False` to prepare
            the cache content at initialization. See: :py:class:`monai.data.CacheDataset`.

    Raises:
        ValueError: When ``root_dir`` is not a directory.
        ValueError: When ``task`` is not one of ["Task01_BrainTumour", "Task02_Heart",
            "Task03_Liver", "Task04_Hippocampus", "Task05_Prostate", "Task06_Lung", "Task07_Pancreas",
            "Task08_HepaticVessel", "Task09_Spleen", "Task10_Colon"].
        RuntimeError: When ``dataset_dir`` doesn't exist and downloading is not selected (``download=False``).

    Example::

        transform = Compose(
            [
                LoadImaged(keys=["image", "label"]),
                EnsureChannelFirstd(keys=["image", "label"]),
                ScaleIntensityd(keys="image"),
                ToTensord(keys=["image", "label"]),
            ]
        )

        val_data = DecathlonDataset(
            root_dir="./", task="Task09_Spleen", transform=transform, section="validation", seed=12345, download=True
        )

        print(val_data[0]["image"], val_data[0]["label"])

    """

    resource = {
        "Task01_BrainTumour": "https://msd-for-monai.s3-us-west-2.amazonaws.com/Task01_BrainTumour.tar",
        "Task02_Heart": "https://msd-for-monai.s3-us-west-2.amazonaws.com/Task02_Heart.tar",
        "Task03_Liver": "https://msd-for-monai.s3-us-west-2.amazonaws.com/Task03_Liver.tar",
        "Task04_Hippocampus": "https://msd-for-monai.s3-us-west-2.amazonaws.com/Task04_Hippocampus.tar",
        "Task05_Prostate": "https://msd-for-monai.s3-us-west-2.amazonaws.com/Task05_Prostate.tar",
        "Task06_Lung": "https://msd-for-monai.s3-us-west-2.amazonaws.com/Task06_Lung.tar",
        "Task07_Pancreas": "https://msd-for-monai.s3-us-west-2.amazonaws.com/Task07_Pancreas.tar",
        "Task08_HepaticVessel": "https://msd-for-monai.s3-us-west-2.amazonaws.com/Task08_HepaticVessel.tar",
        "Task09_Spleen": "https://msd-for-monai.s3-us-west-2.amazonaws.com/Task09_Spleen.tar",
        "Task10_Colon": "https://msd-for-monai.s3-us-west-2.amazonaws.com/Task10_Colon.tar",
    }
    md5 = {
        "Task01_BrainTumour": "240a19d752f0d9e9101544901065d872",
        "Task02_Heart": "06ee59366e1e5124267b774dbd654057",
        "Task03_Liver": "a90ec6c4aa7f6a3d087205e23d4e6397",
        "Task04_Hippocampus": "9d24dba78a72977dbd1d2e110310f31b",
        "Task05_Prostate": "35138f08b1efaef89d7424d2bcc928db",
        "Task06_Lung": "8afd997733c7fc0432f71255ba4e52dc",
        "Task07_Pancreas": "4f7080cfca169fa8066d17ce6eb061e4",
        "Task08_HepaticVessel": "641d79e80ec66453921d997fbf12a29c",
        "Task09_Spleen": "410d4a301da4e5b2f6f86ec3ddba524e",
        "Task10_Colon": "bad7a188931dc2f6acf72b08eb6202d0",
    }

    def __init__(
        self,
        root_dir: PathLike,
        task: str,
        section: str,
        transform: Sequence[Callable] | Callable = (),
        download: bool = False,
        seed: int = 0,
        val_frac: float = 0.2,
        cache_num: int = sys.maxsize,
        cache_rate: float = 1.0,
        num_workers: int = 1,
        progress: bool = True,
        copy_cache: bool = True,
        as_contiguous: bool = True,
        runtime_cache: bool = False,
    ) -> None:
        root_dir = Path(root_dir)
        if not root_dir.is_dir():
            raise ValueError("Root directory root_dir must be a directory.")
        self.section = section
        self.val_frac = val_frac
        self.set_random_state(seed=seed)
        if task not in self.resource:
            raise ValueError(f"Unsupported task: {task}, available options are: {list(self.resource.keys())}.")
        dataset_dir = root_dir / task
        tarfile_name = f"{dataset_dir}.tar"
        if download:
            download_and_extract(
                url=self.resource[task],
                filepath=tarfile_name,
                output_dir=root_dir,
                hash_val=self.md5[task],
                hash_type="md5",
                progress=progress,
            )

        if not dataset_dir.exists():
            raise RuntimeError(
                f"Cannot find dataset directory: {dataset_dir}, please use download=True to download it."
            )
        self.indices: np.ndarray = np.array([])
        data = self._generate_data_list(dataset_dir)
        # as `release` key has typo in Task04 config file, ignore it.
        property_keys = [
            "name",
            "description",
            "reference",
            "licence",
            "tensorImageSize",
            "modality",
            "labels",
            "numTraining",
            "numTest",
        ]
        self._properties = load_decathlon_properties(dataset_dir / "dataset.json", property_keys)
        if transform == ():
            transform = LoadImaged(["image", "label"])
        CacheDataset.__init__(
            self,
            data=data,
            transform=transform,
            cache_num=cache_num,
            cache_rate=cache_rate,
            num_workers=num_workers,
            progress=progress,
            copy_cache=copy_cache,
            as_contiguous=as_contiguous,
            runtime_cache=runtime_cache,
        )

    def get_indices(self) -> np.ndarray:
        """
        Get the indices of datalist used in this dataset.

        """
        return self.indices

    def randomize(self, data: np.ndarray) -> None:
        self.R.shuffle(data)

    def get_properties(self, keys: Sequence[str] | str | None = None) -> dict:
        """
        Get the loaded properties of dataset with specified keys.
        If no keys specified, return all the loaded properties.

        """
        if keys is None:
            return self._properties
        if self._properties is not None:
            return {key: self._properties[key] for key in ensure_tuple(keys)}
        return {}

    def _generate_data_list(self, dataset_dir: PathLike) -> list[dict]:
        # the types of the item in data list should be compatible with the dataloader
        dataset_dir = Path(dataset_dir)
        section = "training" if self.section in ["training", "validation"] else "test"
        datalist = load_decathlon_datalist(dataset_dir / "dataset.json", True, section)
        return self._split_datalist(datalist)

    def _split_datalist(self, datalist: list[dict]) -> list[dict]:
        if self.section == "test":
            return datalist
        length = len(datalist)
        indices = np.arange(length)
        self.randomize(indices)

        val_length = int(length * self.val_frac)
        if self.section == "training":
            self.indices = indices[val_length:]
        else:
            self.indices = indices[:val_length]

        return [datalist[i] for i in self.indices]


class TciaDataset(Randomizable, CacheDataset):
    """
    The Dataset to automatically download the data from a public The Cancer Imaging Archive (TCIA) dataset
    and generate items for training, validation or test.

    The Highdicom library is used to load dicom data with modality "SEG", but only a part of collections are
    supported, such as: "C4KC-KiTS", "NSCLC-Radiomics", "NSCLC-Radiomics-Interobserver1", " QIN-PROSTATE-Repeatability"
    and "PROSTATEx". Therefore, if "seg" is included in `keys` of the `LoadImaged` transform and loading some
    other collections, errors may be raised. For supported collections, the original "SEG" information may not
    always be consistent for each dicom file. Therefore, to avoid creating different format of labels, please use
    the `label_dict` argument of `PydicomReader` when calling the `LoadImaged` transform. The prepared label dicts
    of collections that are mentioned above is also saved in: `monai.apps.tcia.TCIA_LABEL_DICT`. You can also refer
    to the second example bellow.


    This class is based on :py:class:`monai.data.CacheDataset` to accelerate the training process.

    Args:
        root_dir: user's local directory for caching and loading the TCIA dataset.
        collection: name of a TCIA collection.
            a TCIA dataset is defined as a collection. Please check the following list to browse
            the collection list (only public collections can be downloaded):
            https://www.cancerimagingarchive.net/collections/
        section: expected data section, can be: `training`, `validation` or `test`.
        transform: transforms to execute operations on input data.
            for further usage, use `EnsureChannelFirstd` to convert the shape to [C, H, W, D].
            If not specified, `LoadImaged(reader="PydicomReader", keys=["image"])` will be used as the default
            transform. In addition, we suggest to set the argument `labels` for `PydicomReader` if segmentations
            are needed to be loaded. The original labels for each dicom series may be different, using this argument
            is able to unify the format of labels.
        download: whether to download and extract the dataset, default is False.
            if expected file already exists, skip downloading even set it to True.
            user can manually copy tar file or dataset folder to the root directory.
        download_len: number of series that will be downloaded, the value should be larger than 0 or -1, where -1 means
            all series will be downloaded. Default is -1.
        seg_type: modality type of segmentation that is used to do the first step download. Default is "SEG".
        modality_tag: tag of modality. Default is (0x0008, 0x0060).
        ref_series_uid_tag: tag of referenced Series Instance UID. Default is (0x0020, 0x000e).
        ref_sop_uid_tag: tag of referenced SOP Instance UID. Default is (0x0008, 0x1155).
        specific_tags: tags that will be loaded for "SEG" series. This argument will be used in
            `monai.data.PydicomReader`. Default is [(0x0008, 0x1115), (0x0008,0x1140), (0x3006, 0x0010),
            (0x0020,0x000D), (0x0010,0x0010), (0x0010,0x0020), (0x0020,0x0011), (0x0020,0x0012)].
        fname_regex: a regular expression to match the file names when the input is a folder.
            If provided, only the matched files will be included. For example, to include the file name
            "image_0001.dcm", the regular expression could be `".*image_(\\d+).dcm"`.
            Default to `"^(?!.*LICENSE).*"`, ignoring any file name containing `"LICENSE"`.
        val_frac: percentage of validation fraction in the whole dataset, default is 0.2.
        seed: random seed to randomly shuffle the datalist before splitting into training and validation, default is 0.
            note to set same seed for `training` and `validation` sections.
        cache_num: number of items to be cached. Default is `sys.maxsize`.
            will take the minimum of (cache_num, data_length x cache_rate, data_length).
        cache_rate: percentage of cached data in total, default is 0.0 (no cache).
            will take the minimum of (cache_num, data_length x cache_rate, data_length).
        num_workers: the number of worker threads if computing cache in the initialization.
            If num_workers is None then the number returned by os.cpu_count() is used.
            If a value less than 1 is specified, 1 will be used instead.
        progress: whether to display a progress bar when downloading dataset and computing the transform cache content.
        copy_cache: whether to `deepcopy` the cache content before applying the random transforms,
            default to `True`. if the random transforms don't modify the cached content
            (for example, randomly crop from the cached image and deepcopy the crop region)
            or if every cache item is only used once in a `multi-processing` environment,
            may set `copy=False` for better performance.
        as_contiguous: whether to convert the cached NumPy array or PyTorch tensor to be contiguous.
            it may help improve the performance of following logic.
        runtime_cache: whether to compute cache at the runtime, default to `False` to prepare
            the cache content at initialization. See: :py:class:`monai.data.CacheDataset`.

    Example::

        # collection is "Pancreatic-CT-CBCT-SEG", seg_type is "RTSTRUCT"
        data = TciaDataset(
            root_dir="./", collection="Pancreatic-CT-CBCT-SEG", seg_type="RTSTRUCT", download=True
        )

        # collection is "C4KC-KiTS", seg_type is "SEG", and load both images and segmentations
        from monai.apps.tcia import TCIA_LABEL_DICT
        transform = Compose(
            [
                LoadImaged(reader="PydicomReader", keys=["image", "seg"], label_dict=TCIA_LABEL_DICT["C4KC-KiTS"]),
                EnsureChannelFirstd(keys=["image", "seg"]),
                ResampleToMatchd(keys="image", key_dst="seg"),
            ]
        )
        data = TciaDataset(
            root_dir="./", collection="C4KC-KiTS", section="validation", seed=12345, download=True
        )

        print(data[0]["seg"].shape)

    """

    def __init__(
        self,
        root_dir: PathLike,
        collection: str,
        section: str,
        transform: Sequence[Callable] | Callable = (),
        download: bool = False,
        download_len: int = -1,
        seg_type: str = "SEG",
        modality_tag: tuple = (0x0008, 0x0060),
        ref_series_uid_tag: tuple = (0x0020, 0x000E),
        ref_sop_uid_tag: tuple = (0x0008, 0x1155),
        specific_tags: tuple = (
            (0x0008, 0x1115),  # Referenced Series Sequence
            (0x0008, 0x1140),  # Referenced Image Sequence
            (0x3006, 0x0010),  # Referenced Frame of Reference Sequence
            (0x0020, 0x000D),  # Study Instance UID
            (0x0010, 0x0010),  # Patient's Name
            (0x0010, 0x0020),  # Patient ID
            (0x0020, 0x0011),  # Series Number
            (0x0020, 0x0012),  # Acquisition Number
        ),
        fname_regex: str = DCM_FILENAME_REGEX,
        seed: int = 0,
        val_frac: float = 0.2,
        cache_num: int = sys.maxsize,
        cache_rate: float = 0.0,
        num_workers: int = 1,
        progress: bool = True,
        copy_cache: bool = True,
        as_contiguous: bool = True,
        runtime_cache: bool = False,
    ) -> None:
        root_dir = Path(root_dir)
        if not root_dir.is_dir():
            raise ValueError("Root directory root_dir must be a directory.")

        self.section = section
        self.val_frac = val_frac
        self.seg_type = seg_type
        self.modality_tag = modality_tag
        self.ref_series_uid_tag = ref_series_uid_tag
        self.ref_sop_uid_tag = ref_sop_uid_tag

        self.set_random_state(seed=seed)
        download_dir = os.path.join(root_dir, collection)
        load_tags = list(specific_tags)
        load_tags += [modality_tag]
        self.load_tags = load_tags
        if download:
            seg_series_list = get_tcia_metadata(
                query=f"getSeries?Collection={collection}&Modality={seg_type}", attribute="SeriesInstanceUID"
            )
            if download_len > 0:
                seg_series_list = seg_series_list[:download_len]
            if len(seg_series_list) == 0:
                raise ValueError(f"Cannot find data with collection: {collection} seg_type: {seg_type}")
            for series_uid in seg_series_list:
                self._download_series_reference_data(series_uid, download_dir)

        if not os.path.exists(download_dir):
            raise RuntimeError(f"Cannot find dataset directory: {download_dir}.")
        self.fname_regex = fname_regex

        self.indices: np.ndarray = np.array([])
        self.datalist = self._generate_data_list(download_dir)

        if transform == ():
            transform = LoadImaged(keys=["image"], reader="PydicomReader", fname_regex=self.fname_regex)
        CacheDataset.__init__(
            self,
            data=self.datalist,
            transform=transform,
            cache_num=cache_num,
            cache_rate=cache_rate,
            num_workers=num_workers,
            progress=progress,
            copy_cache=copy_cache,
            as_contiguous=as_contiguous,
            runtime_cache=runtime_cache,
        )

    def get_indices(self) -> np.ndarray:
        """
        Get the indices of datalist used in this dataset.

        """
        return self.indices

    def randomize(self, data: np.ndarray) -> None:
        self.R.shuffle(data)

    def _download_series_reference_data(self, series_uid: str, download_dir: str) -> None:
        """
        First of all, download a series from TCIA according to `series_uid`.
        Then find all referenced series and download.
        """
        seg_first_dir = os.path.join(download_dir, "raw", series_uid)
        download_tcia_series_instance(
            series_uid=series_uid, download_dir=download_dir, output_dir=seg_first_dir, check_md5=False
        )
        dicom_files = [f for f in sorted(os.listdir(seg_first_dir)) if f.endswith(".dcm")]
        # achieve series number and patient id from the first dicom file
        dcm_path = os.path.join(seg_first_dir, dicom_files[0])
        ds = PydicomReader(stop_before_pixels=True, specific_tags=self.load_tags).read(dcm_path)
        # (0x0010,0x0020) and (0x0010,0x0010), better to be contained in `specific_tags`
        patient_id = ds.PatientID if ds.PatientID else ds.PatientName
        if not patient_id:
            warnings.warn(f"unable to find patient name of dicom file: {dcm_path}, use 'patient' instead.")
            patient_id = "patient"
        # (0x0020,0x0011) and (0x0020,0x0012), better to be contained in `specific_tags`
        series_num = ds.SeriesNumber if ds.SeriesNumber else ds.AcquisitionNumber
        if not series_num:
            warnings.warn(f"unable to find series number of dicom file: {dcm_path}, use '0' instead.")
            series_num = 0

        series_num = str(series_num)
        seg_dir = os.path.join(download_dir, patient_id, series_num, self.seg_type.lower())
        dcm_dir = os.path.join(download_dir, patient_id, series_num, "image")

        # get ref uuid
        ref_uid_list = []
        for dcm_file in dicom_files:
            dcm_path = os.path.join(seg_first_dir, dcm_file)
            ds = PydicomReader(stop_before_pixels=True, specific_tags=self.load_tags).read(dcm_path)
            if ds[self.modality_tag].value == self.seg_type:
                ref_uid = get_tcia_ref_uid(
                    ds, find_sop=False, ref_series_uid_tag=self.ref_series_uid_tag, ref_sop_uid_tag=self.ref_sop_uid_tag
                )
                if ref_uid == "":
                    ref_sop_uid = get_tcia_ref_uid(
                        ds,
                        find_sop=True,
                        ref_series_uid_tag=self.ref_series_uid_tag,
                        ref_sop_uid_tag=self.ref_sop_uid_tag,
                    )
                    ref_uid = match_tcia_ref_uid_in_study(ds.StudyInstanceUID, ref_sop_uid)
                if ref_uid != "":
                    ref_uid_list.append(ref_uid)
        if not ref_uid_list:
            warnings.warn(f"Cannot find the referenced Series Instance UID from series: {series_uid}.")
        else:
            download_tcia_series_instance(
                series_uid=ref_uid_list[0], download_dir=download_dir, output_dir=dcm_dir, check_md5=False
            )
        if not os.path.exists(seg_dir):
            shutil.copytree(seg_first_dir, seg_dir)

    def _generate_data_list(self, dataset_dir: PathLike) -> list[dict]:
        # the types of the item in data list should be compatible with the dataloader
        dataset_dir = Path(dataset_dir)
        datalist = []
        patient_list = [f.name for f in os.scandir(dataset_dir) if f.is_dir() and f.name != "raw"]
        for patient_id in patient_list:
            series_list = [f.name for f in os.scandir(os.path.join(dataset_dir, patient_id)) if f.is_dir()]
            for series_num in series_list:
                seg_key = self.seg_type.lower()
                image_path = os.path.join(dataset_dir, patient_id, series_num, "image")
                mask_path = os.path.join(dataset_dir, patient_id, series_num, seg_key)

                if os.path.exists(image_path):
                    datalist.append({"image": image_path, seg_key: mask_path})
                else:
                    datalist.append({seg_key: mask_path})

        return self._split_datalist(datalist)

    def _split_datalist(self, datalist: list[dict]) -> list[dict]:
        if self.section == "test":
            return datalist
        length = len(datalist)
        indices = np.arange(length)
        self.randomize(indices)

        val_length = int(length * self.val_frac)
        if self.section == "training":
            self.indices = indices[val_length:]
        else:
            self.indices = indices[:val_length]

        return [datalist[i] for i in self.indices]


class CrossValidation:
    """
    Cross validation dataset based on the general dataset which must have `_split_datalist` API.

    Args:
        dataset_cls: dataset class to be used to create the cross validation partitions.
            It must have `_split_datalist` API.
        nfolds: number of folds to split the data for cross validation.
        seed: random seed to randomly shuffle the datalist before splitting into N folds, default is 0.
        dataset_params: other additional parameters for the dataset_cls base class.

    Example of 5 folds cross validation training::

        cvdataset = CrossValidation(
            dataset_cls=DecathlonDataset,
            nfolds=5,
            seed=12345,
            root_dir="./",
            task="Task09_Spleen",
            section="training",
            transform=train_transform,
            download=True,
        )
        dataset_fold0_train = cvdataset.get_dataset(folds=[1, 2, 3, 4])
        dataset_fold0_val = cvdataset.get_dataset(folds=0, transform=val_transform, download=False)
        # execute training for fold 0 ...

        dataset_fold1_train = cvdataset.get_dataset(folds=[0, 2, 3, 4])
        dataset_fold1_val = cvdataset.get_dataset(folds=1, transform=val_transform, download=False)
        # execute training for fold 1 ...

        ...

        dataset_fold4_train = ...
        # execute training for fold 4 ...

    """

    def __init__(self, dataset_cls: object, nfolds: int = 5, seed: int = 0, **dataset_params: Any) -> None:
        if not hasattr(dataset_cls, "_split_datalist"):
            raise ValueError("dataset class must have _split_datalist API.")
        self.dataset_cls = dataset_cls
        self.nfolds = nfolds
        self.seed = seed
        self.dataset_params = dataset_params

    def get_dataset(self, folds: Sequence[int] | int, **dataset_params: Any) -> object:
        """
        Generate dataset based on the specified fold indices in the cross validation group.

        Args:
            folds: index of folds for training or validation, if a list of values, concatenate the data.
            dataset_params: other additional parameters for the dataset_cls base class, will override
                the same parameters in `self.dataset_params`.

        """
        nfolds = self.nfolds
        seed = self.seed
        dataset_params_ = dict(self.dataset_params)
        dataset_params_.update(dataset_params)

        class _NsplitsDataset(self.dataset_cls):  # type: ignore

            def _split_datalist(self, datalist: list[dict]) -> list[dict]:
                data = partition_dataset(data=datalist, num_partitions=nfolds, shuffle=True, seed=seed)
                return select_cross_validation_folds(partitions=data, folds=folds)

        return _NsplitsDataset(**dataset_params_)