Add task conversion for FLAIR images in WMH challenge dataset. (#84)

* add WMH task conversion prototype. Not tested.

* Fix small mistake. The task conversion now runs.

* Black formatting

* Lint with flake8

yucca/task_conversion/Task006_WMH_Flair.py

@@ -0,0 +1,81 @@
+import shutil
+import gzip
+from batchgenerators.utilities.file_and_folder_operations import join, maybe_mkdir_p, subfolders
+from yucca.task_conversion.utils import generate_dataset_json
+from yucca.paths import yucca_raw_data
+
+
+def convert(path: str, subdir: str = "WMH"):
+    """INPUT DATA - Define input path and suffixes"""
+    path = join(path, subdir)
+
+    """ OUTPUT DATA - Define the task name and prefix """
+    task_name = "Task006_WMH_Flair"
+    task_prefix = "WMH_F"
+
+    datasets = ["Amsterdam", "Singapore", "Utrecht"]
+    site = ""
+
+    # Target paths
+    target_base = join(yucca_raw_data, task_name)
+
+    target_imagesTr = join(target_base, "imagesTr")
+    target_labelsTr = join(target_base, "labelsTr")
+
+    target_imagesTs = join(target_base, "imagesTs")
+    target_labelsTs = join(target_base, "labelsTs")
+
+    maybe_mkdir_p(target_imagesTr)
+    maybe_mkdir_p(target_labelsTs)
+    maybe_mkdir_p(target_imagesTs)
+    maybe_mkdir_p(target_labelsTr)
+
+    ###Populate Target Directory###
+    for dataset in datasets:
+        dataset_path = join(path, dataset)
+        if dataset == "Amsterdam":
+            tr_folder = "Train_GE3T"
+        else:
+            tr_folder = "Train"
+
+        train_folder = join(dataset_path, tr_folder)
+        test_folder = join(dataset_path, "Test")
+
+        training_samples = subfolders(train_folder, join=False)
+        test_samples = subfolders(test_folder, join=False)
+
+        # First we sort the training data
+        for sTr in training_samples:
+            # Loading relevant modalities and the ground truth
+            image_file = open(join(train_folder, sTr, "pre", "FLAIR.nii.gz"), "rb")
+            label = open(join(train_folder, sTr, "pre", "wmh.nii.gz"), "rb")
+            shutil.copyfileobj(image_file, gzip.open(f"{target_imagesTr}/{task_prefix}_{sTr}_000.nii.gz", "wb"))
+            shutil.copyfileobj(label, gzip.open(f"{target_labelsTr}/{task_prefix}_{sTr}.nii.gz", "wb"))
+
+        # Now we sort the test data
+        if dataset == "Amsterdam":
+            for site in test_samples:
+                samples = subfolders(join(test_folder, site), join=False)
+                for sTs in samples:
+                    image_file = open(join(test_folder, site, sTs, "pre", "FLAIR.nii.gz"), "rb")
+                    label = open(join(test_folder, site, sTs, "pre", "wmh.nii.gz"), "rb")
+                    shutil.copyfileobj(image_file, gzip.open(f"{target_imagesTs}/{task_prefix}_{sTs}_000.nii.gz", "wb"))
+                    shutil.copyfileobj(label, gzip.open(f"{target_labelsTs}/{task_prefix}_{sTs}.nii.gz", "wb"))
+        else:
+            for sTs in test_samples:
+                image_file = open(join(test_folder, sTs, "pre", "FLAIR.nii.gz"), "rb")
+                label = open(join(test_folder, sTs, "pre", "wmh.nii.gz"), "rb")
+                shutil.copyfileobj(image_file, gzip.open(f"{target_imagesTs}/{task_prefix}_{sTs}_000.nii.gz", "wb"))
+                shutil.copyfileobj(label, gzip.open(f"{target_labelsTs}/{task_prefix}_{sTs}.nii.gz", "wb"))
+
+    generate_dataset_json(
+        join(target_base, "dataset.json"),
+        target_imagesTr,
+        target_imagesTs,
+        ("Flair"),
+        labels={0: "background", 1: "WMH", 2: "Other Pathology"},
+        dataset_name=task_name,
+        license="CC BY-NC 4.0 DEED",
+        dataset_description="White Matter Hyperintensity Segmentation Challenge. Flair images only!",
+        dataset_reference="https://dataverse.nl/dataset.xhtml?persistentId=doi:10.34894/AECRSD",
+    )