Improve dataset card: Add task category, paper/code links, and LMDB sample usage

README.md

@@ -1,16 +1,21 @@
 ---
-license: apache-2.0
 language:
 - en
+license: apache-2.0
+size_categories:
+- 100B<n<1T
 tags:
 - medical
 - pathology
-size_categories:
-- 100B<n<1T
+task_categories:
+- image-feature-extraction
 ---
 
 # CPathPatchFeature: Pre-extracted WSI Features for Computational Pathology
 
+Paper: [Revisiting End-to-End Learning with Slide-level Supervision in Computational Pathology](https://huggingface.co/papers/2506.02408)
+Code: [https://github.com/DearCaat/E2E-WSI-ABMILX](https://github.com/DearCaat/E2E-WSI-ABMILX)
+
 ## Dataset Summary
 
 This dataset provides a comprehensive collection of pre-extracted features from Whole Slide Images (WSIs) for various cancer types, designed to facilitate research in computational pathology. The features are extracted using multiple state-of-the-art encoders, offering a rich resource for developing and evaluating Multiple Instance Learning (MIL) models and other deep learning architectures.
@@ -73,6 +78,58 @@ Then, clone the dataset repository:
 git clone https://huggingface.co/datasets/Dearcat/CPathPatchFeature
 ```
 
+## Sample Usage
+
+Here is an example of loading data from an LMDB dataset, as provided in the GitHub repository:
+
+```python
+import lmdb
+import torch
+import pickle
+from datasets.utils import imfrombytes # Ensure this utility function is correctly referenced
+
+slide_name = "xxxx"  # Example slide name
+path_to_lmdb = "YOUR_PATH_TO_LMDB_FILE" # e.g., "/path/to/my_dataset_256_level0.lmdb"
+
+# Open LMDB dataset
+env = lmdb.open(path_to_lmdb, subdir=False, readonly=True, lock=False,
+                readahead=False, meminit=False, map_size=100 * (1024**3))
+
+with env.begin(write=False) as txn:
+    # Get patch count for the slide
+    pn_dict = pickle.loads(txn.get(b'__pn__'))
+    if slide_name not in pn_dict:
+        raise ValueError(f"Slide ID {slide_name} not found in LMDB metadata.")
+    num_patches = pn_dict[slide_name]
+
+    # Generate patch IDs
+    patch_ids = [f"{slide_name}-{i}" for i in range(num_patches)]
+
+    # Allocate memory for patches (adjust dimensions and dtype as needed)
+    # Assuming patches are 224x224, 3 channels, and will be normalized later
+    patches_tensor = torch.empty((len(patch_ids), 3, 224, 224), dtype=torch.float32)
+
+    # Load and decode data into torch.tensor
+    for i, key_str in enumerate(patch_ids):
+        patch_bytes = txn.get(key_str.encode('ascii'))
+        if patch_bytes is None:
+            print(f"Warning: Key {key_str} not found in LMDB.")
+            continue
+        # Assuming the stored value is pickled image bytes
+        img_array = imfrombytes(pickle.loads(patch_bytes).tobytes()) # Or .tobytes() if it's already bytes
+        patches_tensor[i] = torch.from_numpy(img_array.transpose(2, 0, 1)) # HWC to CHW
+
+# Normalize the data (example using ImageNet stats)
+# Ensure values are in [0, 255] before this normalization if they aren't already
+mean = torch.tensor([0.485, 0.456, 0.406]).view((1, 3, 1, 1)) * 255.0
+std = torch.tensor([0.229, 0.224, 0.225]).view((1, 3, 1, 1)) * 255.0
+
+# If your patches_tensor is already in [0,1] range, remove * 255.0 from mean/std
+# If your patches_tensor is uint8 [0,255], convert to float first: patches_tensor.float()
+patches_tensor = (patches_tensor.float() - mean) / std
+
+env.close()
+```
 
 ### Citation
 This dataset has been used in the following publications. If you find it useful for your research, please consider citing them: