Update dataset card: schema, examples, joins

README.md

@@ -1,80 +1,71 @@
 ---
-pretty_name: MultiCaRe Images (metadata)
+pretty_name: MultiCaRe Images
 license: cc-by-4.0
 task_categories:
-- image-classification
-- image-to-text
+  - image-classification
+  - image-to-text
 language:
-- en
+  - en
 size_categories:
-- 100K<n<1M
-dataset_info:
-  features:
-  - name: file_id
-    dtype: string
-  - name: file
-    dtype: string
-  - name: main_image
-    dtype: string
-  - name: image_component
-    dtype: string
-  - name: patient_id
-    dtype: string
-  - name: license
-    dtype: string
-  - name: caption
-    dtype: string
-  - name: case_substring
-    sequence: string
-  - name: image_type
-    dtype: string
-  - name: image_subtype
-    dtype: string
-  - name: radiology_region
-    dtype: string
-  - name: radiology_region_granular
-    dtype: string
-  - name: radiology_view
-    dtype: string
-  - name: image
-    dtype: image
-  - name: labels
-    sequence: string
-  - name: semi_labels
-    sequence: string
-  splits:
-  - name: train
-    num_bytes: 2956191106.45
-    num_examples: 161702
-  download_size: 3238623374
-  dataset_size: 2956191106.45
-configs:
-- config_name: default
-  data_files:
-  - split: train
-    path: data/train-*
+  - 100K<n<1M
 ---
 
-# MultiCaRe Images (metadata)
+# MultiCaRe Images (with pixels)
 
-Per-image metadata, captions, and labels from MultiCaRe (Zenodo 13936721), sourced from `captions_and_labels.csv`.
+Per-image dataset with the actual images, captions, labels, and core metadata from MultiCaRe (Zenodo 13936721). Images are stored inside the dataset shards so you can just load and use.
 
-Schema highlights:
-- file_id: unique image row ID
-- file: image filename; archive = `PMC{file[:4][-1]}.zip`; inner path = `f"{file[:4]}/{file[:5]}/{file}"`
-- main_image: original figure ID (equals `case_images.image_id`)
-- patient_id: equals `cases.case_id`
-- caption: image caption (full or segment)
-- image_type / image_subtype / radiology_*: multiclass columns
-- labels: supervised multilabels (string list)
-- semi_labels: semi-supervised multilabels (string list, incomplete by design)
+Highlights
+- 161k+ images across radiology, pathology, endoscopy, medical photographs, ophthalmic imaging, electrography, and charts.
+- Supervised multilabels (89-class reduced taxonomy) and optional semi-supervised labels.
+- Stable join keys to link with cases and articles datasets.
+
+Schema
+- file_id: unique row ID for the processed image file
+- image: datasets.Image (PIL-compatible)
+- file: processed image filename
+- main_image: original figure ID (group identifier for subimages)
+- image_component: subimage reference (e.g. undivided, a, b, …)
+- caption: figure caption (full or segment)
+- labels: list of labels for supervised training (strings)
+- semi_labels: additional labels from the full taxonomy (strings; sparse)
+- image_type, image_subtype, radiology_region, radiology_region_granular, radiology_view: multiclass attributes
+- patient_id: case identifier (join to cases.case_id)
 - license: per-article OA license
 
-Join keys to other repos:
-- `patient_id` ↔ `case_id` (cases)
-- `main_image` ↔ `image_id` (case-images)
-- via cases → `article_id` (articles, abstracts)
+Quick start
+```python
+from datasets import load_dataset
+
+tok = "<your HF token>"  # or set HF_TOKEN env var
+ds = load_dataset("MaziyarPanahi/multicare-images", split="train", use_auth_token=tok)
+
+img = ds[0]["image"]        # PIL.Image.Image
+caption = ds[0]["caption"]
+labels = ds[0]["labels"]     # list[str]
+img.show()
+```
+
+Join examples
+```python
+from datasets import load_dataset
+tok = "<token>"
+img = load_dataset("MaziyarPanahi/multicare-images", split="train", use_auth_token=tok)
+cas = load_dataset("MaziyarPanahi/multicare-cases", split="train", use_auth_token=tok)
+art = load_dataset("MaziyarPanahi/multicare-articles", split="train", use_auth_token=tok)
+
+# Example: fetch one case and its first image
+case = cas[0]
+case_id = case["case_id"]
+imgs_for_case = img.filter(lambda e: e["patient_id"] == case_id)
+print(case["case_text"][:400])
+imgs_for_case[0]["image"].show()
+```
+
+Splitting tips
+- Avoid leakage by splitting at patient_id (case) or article_id level.
 
-Note: This repo does not host the actual images. Use your local download (or Zenodo) and map
-`file` to the right zip and inner path as described above.
+License
+- The dataset is CC-BY-4.0. Each item also retains the per-article OA license string.
 
+Cite
+- DOI: 10.5281/zenodo.13936721