Rename CRC-HE to CRC (#189)

README.md

@@ -46,7 +46,7 @@ The following datasets are supported
 
 ### Vision
 - [BACH](./docs/datasets/bach.md)
-- [CRC-HE](./docs/datasets/crc_he.md)
+- [CRC](./docs/datasets/crc.md)
 - [PatchCamelyon](./docs/datasets/patch_camelyon.md)
 - [TotalSegmentator](./docs/datasets/total_segmentator.md)
 

configs/vision/dino_vits16/offline/crc_he_nonorm.yaml → configs/vision/dino_vits16/offline/crc.yaml

@@ -2,7 +2,7 @@
 trainer:
   class_path: eva.Trainer
   init_args:
-    default_root_dir: &LIGHTNING_ROOT ${oc.env:LIGHTNING_ROOT, logs/offline/crc_he_nonorm}
+    default_root_dir: &LIGHTNING_ROOT ${oc.env:LIGHTNING_ROOT, logs/dino_vits16/offline/crc}
     max_steps: &MAX_STEPS 12500
     callbacks:
       - class_path: pytorch_lightning.callbacks.LearningRateMonitor
@@ -23,7 +23,7 @@ trainer:
           mode: *MONITOR_METRIC_MODE
       - class_path: eva.callbacks.EmbeddingsWriter
         init_args:
-          output_dir: &EMBEDDINGS_DIR ${oc.env:EMBEDDINGS_ROOT, ./data/embeddings}/crc_he_nonorm
+          output_dir: &EMBEDDINGS_DIR ${oc.env:EMBEDDINGS_ROOT, ./data/embeddings}/crc
           dataloader_idx_map:
             0: train
             1: val
@@ -83,17 +83,17 @@ data:
           <<: *DATASET_ARGS
           split: val
       predict:
-        - class_path: eva.vision.datasets.CRC_HE_NONORM
+        - class_path: eva.vision.datasets.CRC
           init_args: &PREDICT_DATASET_ARGS
-            root: ${oc.env:DATA_ROOT, ./data}/crc_he
+            root: ${oc.env:DATA_ROOT, ./data}/crc
             split: train
             download: ${oc.env:DOWNLOAD_DATA, true}
             image_transforms:
               class_path: eva.vision.data.transforms.common.ResizeAndCrop
               init_args:
                 mean: [0.485, 0.456, 0.406]
                 std: [0.229, 0.224, 0.225]
-        - class_path: eva.vision.datasets.CRC_HE_NONORM
+        - class_path: eva.vision.datasets.CRC
           init_args:
             <<: *PREDICT_DATASET_ARGS
             split: val

configs/vision/dino_vits16/offline/crc_he.yaml → configs/vision/dino_vits16/offline/crc_nonorm.yaml

@@ -2,7 +2,7 @@
 trainer:
   class_path: eva.Trainer
   init_args:
-    default_root_dir: &LIGHTNING_ROOT ${oc.env:LIGHTNING_ROOT, logs/offline/crc_he}
+    default_root_dir: &LIGHTNING_ROOT ${oc.env:LIGHTNING_ROOT, logs/dino_vits16/offline/crc_nonorm}
     max_steps: &MAX_STEPS 12500
     callbacks:
       - class_path: pytorch_lightning.callbacks.LearningRateMonitor
@@ -23,7 +23,7 @@ trainer:
           mode: *MONITOR_METRIC_MODE
       - class_path: eva.callbacks.EmbeddingsWriter
         init_args:
-          output_dir: &EMBEDDINGS_DIR ${oc.env:EMBEDDINGS_ROOT, ./data/embeddings}/crc_he
+          output_dir: &EMBEDDINGS_DIR ${oc.env:EMBEDDINGS_ROOT, ./data/embeddings}/crc_nonorm
           dataloader_idx_map:
             0: train
             1: val
@@ -83,7 +83,7 @@ data:
           <<: *DATASET_ARGS
           split: val
       predict:
-        - class_path: eva.vision.datasets.CRC_HE
+        - class_path: eva.vision.datasets.CRC_NONORM
           init_args: &PREDICT_DATASET_ARGS
             root: ${oc.env:DATA_ROOT, ./data}/crc_he
             split: train
@@ -93,7 +93,7 @@ data:
               init_args:
                 mean: [0.485, 0.456, 0.406]
                 std: [0.229, 0.224, 0.225]
-        - class_path: eva.vision.datasets.CRC_HE
+        - class_path: eva.vision.datasets.CRC_NONORM
           init_args:
             <<: *PREDICT_DATASET_ARGS
             split: val

configs/vision/dino_vits16/online/crc_he.yaml → configs/vision/dino_vits16/online/crc.yaml

@@ -2,7 +2,7 @@
 trainer:
   class_path: eva.Trainer
   init_args:
-    default_root_dir: &LIGHTNING_ROOT ${oc.env:LIGHTNING_ROOT, logs/online/crc_he}
+    default_root_dir: &LIGHTNING_ROOT ${oc.env:LIGHTNING_ROOT, logs/dino_vits16/online/crc}
     max_steps: &MAX_STEPS 12500
     callbacks:
       - class_path: pytorch_lightning.callbacks.LearningRateMonitor
@@ -65,9 +65,9 @@ data:
   init_args:
     datasets:
       train:
-        class_path: eva.vision.datasets.CRC_HE
+        class_path: eva.vision.datasets.CRC
         init_args: &DATASET_ARGS
-          root: ${oc.env:DATA_ROOT, ./data}/crc_he
+          root: ${oc.env:DATA_ROOT, ./data}/crc
           split: train
           download: ${oc.env:DOWNLOAD_DATA, true}
           image_transforms:
@@ -76,7 +76,7 @@ data:
               mean: [0.485, 0.456, 0.406]
               std: [0.229, 0.224, 0.225]
       val:
-        class_path: eva.vision.datasets.CRC_HE
+        class_path: eva.vision.datasets.CRC
         init_args:
           <<: *DATASET_ARGS
           split: val

configs/vision/dino_vits16/online/crc_he_nonorm.yaml → configs/vision/dino_vits16/online/crc_nonorm.yaml

@@ -2,7 +2,7 @@
 trainer:
   class_path: eva.Trainer
   init_args:
-    default_root_dir: &LIGHTNING_ROOT ${oc.env:LIGHTNING_ROOT, logs/offline/crc_he}
+    default_root_dir: &LIGHTNING_ROOT ${oc.env:LIGHTNING_ROOT, logs/dino_vits16/online/crc_nonorm}
     max_steps: &MAX_STEPS 12500
     callbacks:
       - class_path: pytorch_lightning.callbacks.LearningRateMonitor
@@ -65,9 +65,9 @@ data:
   init_args:
     datasets:
       train:
-        class_path: eva.vision.datasets.CRC_HE_NONORM
+        class_path: eva.vision.datasets.CRC_NONORM
         init_args: &DATASET_ARGS
-          root: ${oc.env:DATA_ROOT, ./data}/crc_he
+          root: ${oc.env:DATA_ROOT, ./data}/crc
           split: train
           download: ${oc.env:DOWNLOAD_DATA, true}
           image_transforms:
@@ -76,7 +76,7 @@ data:
               mean: [0.485, 0.456, 0.406]
               std: [0.229, 0.224, 0.225]
       val:
-        class_path: eva.vision.datasets.CRC_HE_NONORM
+        class_path: eva.vision.datasets.CRC_NONORM
         init_args:
           <<: *DATASET_ARGS
           split: val

docs/datasets/crc_he.md → docs/datasets/crc.md

@@ -1,4 +1,4 @@
-# CRC-HE
+# CRC
 
 The CRC-HE dataset consists of labelled patches (9 classes) from colorectal cancer (CRC) and normal tissue. We use the `NCT-CRC-HE-100K` dataset for training and validation and the `CRC-VAL-HE-7K for testing`. 
 
@@ -19,19 +19,19 @@ The tissue classes are: Adipose (ADI), background (BACK), debris (DEB), lymphocy
 | **Image dimension**  | 224 x 224 x 3                                       |
 | **FoV (μm/px)**      | 20x (0.5)                                           |
 | **Files format**     | `.tif` images                                       |
-| **Number of images** | 107,180 (100k train, 7.2k val)                  |
-| **Splits in use**    | NCT-CRC-HE-100K-NONORM (train), CRC-VAL-HE-7K (val) |
+| **Number of images** | 107,180 (100k train, 7.2k val)                      |
+| **Splits in use**    | NCT-CRC-HE-100K (train), CRC-VAL-HE-7K (val)        |
 
 
 ### Splits
 
 We use the splits according to the data sources:
 
- - Train split: `NCT-CRC-HE-100K-NONORM`
+ - Train split: `NCT-CRC-HE-100K`
  - Validation split: `CRC-VAL-HE-7K`
 
-| Splits | Train           | Validation   | 
-|---|-----------------|--------------|
+| Splits   | Train           | Validation   | 
+|----------|-----------------|--------------|
 | #Samples | 100,000 (93.3%) | 7,180 (6.7%) | 
 
 A test split is not provided. Because the patient information for the training data is not available, dividing the 
@@ -40,9 +40,18 @@ __eva__ therefore reports evaluation results for CRC HE on the validation split.
 
 ### Organization
 
-The data `NCT-CRC-HE-100K-NONORM.zip` and `CRC-VAL-HE-7K.zip` from [zenodo](https://zenodo.org/records/1214456) are organized as follows:
+The data `NCT-CRC-HE-100K.zip`, `NCT-CRC-HE-100K-NONORM.zip` and `CRC-VAL-HE-7K.zip`
+from [zenodo](https://zenodo.org/records/1214456) are organized as follows:
 
 ```
+NCT-CRC-HE-100K                # All images used for training
+├── ADI                        # All labelled patches belonging to the 1st class
+│   ├── ADI-AAAFLCLY.tif
+│   ├── ...
+├── BACK                       # All labelled patches belonging to the 2nd class
+│   ├── ...
+└── ...
+
 NCT-CRC-HE-100K-NONORM         # All images used for training
 ├── ADI                        # All labelled patches belonging to the 1st class
 │   ├── ADI-AAAFLCLY.tif
@@ -58,12 +67,12 @@ CRC-VAL-HE-7K                  # All images used for validation
 
 ## Download and preprocessing
 
-The `CRC_HE` dataset class supports download the data no runtime with the initialized argument
+The `CRC` dataset class supports download the data no runtime with the initialized argument
 `download: bool = True`.
 
 ## Relevant links
 
-* [CRC-HE datasets on zenodo](https://zenodo.org/records/1214456)
+* [CRC datasets on zenodo](https://zenodo.org/records/1214456)
 * [Reference API Vision dataset classes](../reference/vision/data/datasets.md)
 
 

docs/datasets/index.md

@@ -1,16 +1,16 @@
 # Datasets
 
-**eva** provides out-of-the-box support for several public datasets. When possible, the corresponding dataset classes facilitate automatic download to disk, if not possible, this documentation provides download instructions.
+**eva** provides native support for several public datasets. When possible, the corresponding dataset classes facilitate automatic download to disk, if not possible, this documentation provides download instructions.
 
 ## Vision Datasets Overview
 
 ### Whole Slide (WSI) and microscopy image datasets
 
-| Dataset                            | #Patches | Patch Size | FoV (μm/px) | Task                       | Cancer Type | Download provided
-|------------------------------------|----------|------------|---|----------------------------|-------------| ---| 
-| [BACH](bach.md)                    | 400      | 2048x1536  | 20x (0.5) | Classification (4 classes) | Breast      | Yes |
-| [CRC HE](crc_he.md)                | 107,180  | 224x224    | 20x (0.5) | Classification (9 classes) | Colorectal  | Yes |
-| [PatchCamelyon](patch_camelyon.md) | 327,680  | 96x96      | 10x (1.0) \* | Classification (2 classes) | Breast      | Yes |
+| Dataset                            | #Patches | Patch Size | FoV (μm/px)  | Task                       | Cancer Type |
+|------------------------------------|----------|------------|--------------|----------------------------|-------------|
+| [BACH](bach.md)                    | 400      | 2048x1536  | 20x (0.5)    | Classification (4 classes) | Breast      |
+| [CRC](crc.md)                      | 107,180  | 224x224    | 20x (0.5)    | Classification (9 classes) | Colorectal  |
+| [PatchCamelyon](patch_camelyon.md) | 327,680  | 96x96      | 10x (1.0) \* | Classification (2 classes) | Breast      |
 
 \* The slides were acquired and digitized at 2 different centres using a 40x objective but under-sampled to 10x to increase the field of view. Some papers do categorize it as 10x.
 

src/eva/vision/data/datasets/__init__.py

@@ -2,8 +2,8 @@
 
 from eva.vision.data.datasets.classification import (
     BACH,
-    CRC_HE,
-    CRC_HE_NONORM,
+    CRC,
+    CRC_NONORM,
     PatchCamelyon,
     TotalSegmentatorClassification,
 )
@@ -13,8 +13,8 @@
 
 __all__ = [
     "BACH",
-    "CRC_HE",
-    "CRC_HE_NONORM",
+    "CRC",
+    "CRC_NONORM",
     "PatchEmbeddingDataset",
     "ImageSegmentation",
     "SlideEmbeddingDataset",

src/eva/vision/data/datasets/classification/__init__.py

@@ -1,9 +1,15 @@
 """Image classification datasets API."""
 
 from eva.vision.data.datasets.classification.bach import BACH
-from eva.vision.data.datasets.classification.crc_he import CRC_HE
-from eva.vision.data.datasets.classification.crc_he_nonorm import CRC_HE_NONORM
+from eva.vision.data.datasets.classification.crc import CRC
+from eva.vision.data.datasets.classification.crc_nonorm import CRC_NONORM
 from eva.vision.data.datasets.classification.patch_camelyon import PatchCamelyon
 from eva.vision.data.datasets.classification.total_segmentator import TotalSegmentatorClassification
 
-__all__ = ["BACH", "CRC_HE", "CRC_HE_NONORM", "PatchCamelyon", "TotalSegmentatorClassification"]
+__all__ = [
+    "BACH",
+    "CRC",
+    "CRC_NONORM",
+    "PatchCamelyon",
+    "TotalSegmentatorClassification",
+]

src/eva/vision/data/datasets/classification/crc_he.py → src/eva/vision/data/datasets/classification/crc.py

@@ -1,4 +1,4 @@
-"""CRC-HE dataset class."""
+"""CRC dataset class."""
 
 import os
 from typing import Callable, Dict, List, Literal, Tuple
@@ -12,8 +12,8 @@
 from eva.vision.utils import io
 
 
-class CRC_HE(base.ImageClassification):
-    """Dataset class for CRC-HE images and corresponding targets."""
+class CRC(base.ImageClassification):
+    """Dataset class for CRC images and corresponding targets."""
 
     _train_resource: structs.DownloadResource = structs.DownloadResource(
         filename="NCT-CRC-HE-100K.zip",

src/eva/vision/data/datasets/classification/crc_he_nonorm.py → src/eva/vision/data/datasets/classification/crc_nonorm.py

@@ -1,4 +1,4 @@
-"""CRC-HE-NONORM dataset class."""
+"""CRC-NONORM dataset class."""
 
 import os
 from typing import Callable, Dict, List, Literal, Tuple
@@ -12,8 +12,8 @@
 from eva.vision.utils import io
 
 
-class CRC_HE_NONORM(base.ImageClassification):
-    """Dataset class for CRC-HE-NONORM images and corresponding targets."""
+class CRC_NONORM(base.ImageClassification):
+    """Dataset class for CRC-NONORM images and corresponding targets."""
 
     _train_resource: structs.DownloadResource = structs.DownloadResource(
         filename="NCT-CRC-HE-100K-NONORM.zip",

tests/eva/vision/data/datasets/classification/test_crc_he.py → tests/eva/vision/data/datasets/classification/test_crc.py

@@ -1,4 +1,4 @@
-"""CRC_HE dataset tests."""
+"""CRC dataset tests."""
 
 import os
 from typing import Literal
@@ -18,10 +18,10 @@
         ("val", 2),
     ],
 )
-def test_sample(crc_he_dataset: datasets.CRC_HE, index: int) -> None:
+def test_sample(crc_dataset: datasets.CRC, index: int) -> None:
     """Tests the format of a dataset sample."""
     # assert data sample is a tuple
-    sample = crc_he_dataset[index]
+    sample = crc_dataset[index]
     assert isinstance(sample, tuple)
     assert len(sample) == 2
     # assert the format of the `image` and `target`
@@ -33,10 +33,10 @@ def test_sample(crc_he_dataset: datasets.CRC_HE, index: int) -> None:
 
 
 @pytest.fixture(scope="function")
-def crc_he_dataset(split: Literal["train", "val"], assets_path: str) -> datasets.CRC_HE:
-    """CRC_HE dataset fixture."""
-    dataset = datasets.CRC_HE(
-        root=os.path.join(assets_path, "vision", "datasets", "crc_he"),
+def crc_dataset(split: Literal["train", "val"], assets_path: str) -> datasets.CRC:
+    """CRC dataset fixture."""
+    dataset = datasets.CRC(
+        root=os.path.join(assets_path, "vision", "datasets", "crc"),
         split=split,
     )
     dataset.prepare_data()

tests/eva/vision/data/datasets/classification/test_crc_he_nonorm.py → tests/eva/vision/data/datasets/classification/test_crc_nonorm.py

@@ -1,4 +1,4 @@
-"""CRC_HE_NONORM dataset tests."""
+"""CRC_NONORM dataset tests."""
 
 import os
 from typing import Literal
@@ -18,10 +18,10 @@
         ("val", 2),
     ],
 )
-def test_sample(crc_he_nonorm_dataset: datasets.CRC_HE_NONORM, index: int) -> None:
+def test_sample(crc_nonorm_dataset: datasets.CRC_NONORM, index: int) -> None:
     """Tests the format of a dataset sample."""
     # assert data sample is a tuple
-    sample = crc_he_nonorm_dataset[index]
+    sample = crc_nonorm_dataset[index]
     assert isinstance(sample, tuple)
     assert len(sample) == 2
     # assert the format of the `image` and `target`
@@ -33,12 +33,10 @@ def test_sample(crc_he_nonorm_dataset: datasets.CRC_HE_NONORM, index: int) -> No
 
 
 @pytest.fixture(scope="function")
-def crc_he_nonorm_dataset(
-    split: Literal["train", "val"], assets_path: str
-) -> datasets.CRC_HE_NONORM:
+def crc_nonorm_dataset(split: Literal["train", "val"], assets_path: str) -> datasets.CRC_NONORM:
     """CRC_HE_NONORM dataset fixture."""
-    dataset = datasets.CRC_HE_NONORM(
-        root=os.path.join(assets_path, "vision", "datasets", "crc_he"),
+    dataset = datasets.CRC_NONORM(
+        root=os.path.join(assets_path, "vision", "datasets", "crc"),
         split=split,
     )
     dataset.prepare_data()