909 Drop deprecated APIs in IO transforms (#1448)

* [DLMED] remove deprecated APIs in IO transforms

Signed-off-by: Nic Ma <nma@nvidia.com>

docs/source/highlights.md

@@ -39,7 +39,7 @@ There is a rich set of transforms in six categories: Crop & Pad, Intensity, IO,
 ### 2. Medical specific transforms
 MONAI aims at providing a comprehensive medical image specific
 transformations. These currently include, for example:
-- `LoadNifti`:  Load Nifti format file from provided path
+- `LoadImage`:  Load medical specific formats file from provided path
 - `Spacing`:  Resample input image into the specified `pixdim`
 - `Orientation`: Change the image's orientation into the specified `axcodes`
 - `RandGaussianNoise`: Perturb image intensities by adding statistical noises

docs/source/transforms.rst

@@ -231,24 +231,6 @@ IO
     :members:
     :special-members: __call__
 
-`LoadNifti`
-"""""""""""
-.. autoclass:: LoadNifti
-    :members:
-    :special-members: __call__
-
-`LoadPNG`
-"""""""""
-.. autoclass:: LoadPNG
-    :members:
-    :special-members: __call__
-
-`LoadNumpy`
-"""""""""""
-.. autoclass:: LoadNumpy
-    :members:
-    :special-members: __call__
-
 Post-processing
 ^^^^^^^^^^^^^^^
 
@@ -708,36 +690,12 @@ Instensity (Dict)
 IO (Dict)
 ^^^^^^^^^
 
-`LoadDatad`
-"""""""""""
-.. autoclass:: LoadDatad
-    :members:
-    :special-members: __call__
-
 `LoadImaged`
 """"""""""""
 .. autoclass:: LoadImaged
     :members:
     :special-members: __call__
 
-`LoadNiftid`
-""""""""""""
-.. autoclass:: LoadNiftid
-    :members:
-    :special-members: __call__
-
-`LoadPNGd`
-""""""""""
-.. autoclass:: LoadPNGd
-    :members:
-    :special-members: __call__
-
-`LoadNumpyd`
-""""""""""""
-.. autoclass:: LoadNumpyd
-    :members:
-    :special-members: __call__
-
 Post-processing (Dict)
 ^^^^^^^^^^^^^^^^^^^^^^
 

monai/apps/datasets.py

@@ -37,9 +37,7 @@ class MedNISTDataset(Randomizable, CacheDataset):
     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. the default transform is `LoadPNGd`,
-            which can load data into numpy array with [H, W] shape. for further usage, use `AddChanneld`
-            to convert the shape to [C, H, W, D].
+        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.
@@ -158,8 +156,7 @@ class DecathlonDataset(Randomizable, CacheDataset):
             "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. the default transform is `LoadNiftid`,
-            which can load Nifti format data into numpy array with [H, W, D] or [H, W, D, C] shape.
+        transform: transforms to execute operations on input data.
             for further usage, use `AddChanneld` or `AsChannelFirstd` 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.
@@ -185,7 +182,7 @@ class DecathlonDataset(Randomizable, CacheDataset):
 
         transform = Compose(
             [
-                LoadNiftid(keys=["image", "label"]),
+                LoadImaged(keys=["image", "label"]),
                 AddChanneld(keys=["image", "label"]),
                 ScaleIntensityd(keys="image"),
                 ToTensord(keys=["image", "label"]),

monai/data/dataset.py

@@ -89,15 +89,15 @@ class PersistentDataset(Dataset):
 
     .. code-block:: python
 
-        [ LoadNiftid(keys=['image', 'label']),
+        [ LoadImaged(keys=['image', 'label']),
           Orientationd(keys=['image', 'label'], axcodes='RAS'),
           ScaleIntensityRanged(keys=['image'], a_min=-57, a_max=164, b_min=0.0, b_max=1.0, clip=True),
           RandCropByPosNegLabeld(keys=['image', 'label'], label_key='label', spatial_size=(96, 96, 96),
                                  pos=1, neg=1, num_samples=4, image_key='image', image_threshold=0),
           ToTensord(keys=['image', 'label'])]
 
     Upon first use a filename based dataset will be processed by the transform for the
-    [LoadNiftid, Orientationd, ScaleIntensityRanged] and the resulting tensor written to
+    [LoadImaged, Orientationd, ScaleIntensityRanged] and the resulting tensor written to
     the `cache_dir` before applying the remaining random dependant transforms
     [RandCropByPosNegLabeld, ToTensord] elements for use in the analysis.
 
@@ -446,7 +446,7 @@ class CacheDataset(Dataset):
     For example, if the transform is a `Compose` of::
 
         transforms = Compose([
-            LoadNiftid(),
+            LoadImaged(),
             AddChanneld(),
             Spacingd(),
             Orientationd(),
@@ -457,7 +457,7 @@ class CacheDataset(Dataset):
 
     when `transforms` is used in a multi-epoch training pipeline, before the first training epoch,
     this dataset will cache the results up to ``ScaleIntensityRanged``, as
-    all non-random transforms `LoadNiftid`, `AddChanneld`, `Spacingd`, `Orientationd`, `ScaleIntensityRanged`
+    all non-random transforms `LoadImaged`, `AddChanneld`, `Spacingd`, `Orientationd`, `ScaleIntensityRanged`
     can be cached. During training, the dataset will load the cached results and run
     ``RandCropByPosNegLabeld`` and ``ToTensord``, as ``RandCropByPosNegLabeld`` is a randomized transform
     and the outcome not cached.
@@ -825,7 +825,7 @@ class ArrayDataset(Randomizable, _TorchDataset):
 
         img_transform = Compose(
             [
-                LoadNifti(image_only=True),
+                LoadImage(image_only=True),
                 AddChannel(),
                 RandAdjustContrast()
             ]
@@ -834,7 +834,7 @@ class ArrayDataset(Randomizable, _TorchDataset):
 
     If training based on images and the metadata, the array transforms can not be composed
     because several transforms receives multiple parameters or return multiple values. Then Users need
-    to define their own callable method to parse metadata from `LoadNifti` or set `affine` matrix
+    to define their own callable method to parse metadata from `LoadImage` or set `affine` matrix
     to `Spacing` transform::
 
         class TestCompose(Compose):
@@ -845,7 +845,7 @@ def __call__(self, input_):
                 return self.transforms[3](img), metadata
         img_transform = TestCompose(
             [
-                LoadNifti(image_only=False),
+                LoadImage(image_only=False),
                 AddChannel(),
                 Spacing(pixdim=(1.5, 1.5, 3.0)),
                 RandAdjustContrast()

monai/data/nifti_reader.py

@@ -14,7 +14,7 @@
 import numpy as np
 from torch.utils.data import Dataset
 
-from monai.transforms import LoadNifti, Randomizable, apply_transform
+from monai.transforms import LoadImage, Randomizable, apply_transform
 from monai.utils import MAX_SEED, get_seed
 
 
@@ -81,16 +81,19 @@ def randomize(self, data: Optional[Any] = None) -> None:
     def __getitem__(self, index: int):
         self.randomize()
         meta_data = None
-        img_loader = LoadNifti(
-            as_closest_canonical=self.as_closest_canonical, image_only=self.image_only, dtype=self.dtype
+        img_loader = LoadImage(
+            reader="NibabelReader",
+            image_only=self.image_only,
+            dtype=self.dtype,
+            as_closest_canonical=self.as_closest_canonical,
         )
         if self.image_only:
             img = img_loader(self.image_files[index])
         else:
             img, meta_data = img_loader(self.image_files[index])
         seg = None
         if self.seg_files is not None:
-            seg_loader = LoadNifti(image_only=True)
+            seg_loader = LoadImage(image_only=True)
             seg = seg_loader(self.seg_files[index])
         label = None
         if self.labels is not None:

monai/transforms/__init__.py

@@ -137,22 +137,8 @@
     ThresholdIntensityD,
     ThresholdIntensityDict,
 )
-from .io.array import LoadImage, LoadNifti, LoadNumpy, LoadPNG
-from .io.dictionary import (
-    LoadDatad,
-    LoadImaged,
-    LoadImageD,
-    LoadImageDict,
-    LoadNiftid,
-    LoadNiftiD,
-    LoadNiftiDict,
-    LoadNumpyd,
-    LoadNumpyD,
-    LoadNumpyDict,
-    LoadPNGd,
-    LoadPNGD,
-    LoadPNGDict,
-)
+from .io.array import LoadImage
+from .io.dictionary import LoadImaged, LoadImageD, LoadImageDict
 from .post.array import (
     Activations,
     AsDiscrete,

monai/transforms/compose.py

@@ -55,7 +55,7 @@ def __call__(self, data: Any):
         - ``data`` is a Numpy ndarray, PyTorch Tensor or string
         - the data shape can be:
 
-          #. string data without shape, `LoadNifti` and `LoadPNG` transforms expect file paths
+          #. string data without shape, `LoadImage` transform expects file paths
           #. most of the pre-processing transforms expect: ``(num_channels, spatial_dim_1[, spatial_dim_2, ...])``,
              except that `AddChannel` expects (spatial_dim_1[, spatial_dim_2, ...]) and
              `AsChannelFirst` expects (spatial_dim_1[, spatial_dim_2, ...], num_channels)
@@ -282,7 +282,7 @@ def __call__(self, data):
         - ``data[key]`` is a Numpy ndarray, PyTorch Tensor or string, where ``key`` is an element
           of ``self.keys``, the data shape can be:
 
-          #. string data without shape, `LoadNiftid` and `LoadPNGd` transforms expect file paths
+          #. string data without shape, `LoadImaged` transform expects file paths
           #. most of the pre-processing transforms expect: ``(num_channels, spatial_dim_1[, spatial_dim_2, ...])``,
              except that `AddChanneld` expects (spatial_dim_1[, spatial_dim_2, ...]) and
              `AsChannelFirstd` expects (spatial_dim_1[, spatial_dim_2, ...], num_channels)