Create README_MedThinkSeg_BTXRD.md

BTXRD/README_MedThinkSeg_BTXRD.md

@@ -0,0 +1,113 @@
+# BTXRD (Bone Tumor X-ray Radiograph Dataset) — packaged in `saiteja33/MedThink-Seg/BTXRD`
+
+This README documents:
+1) **The original BTXRD dataset** (what it contains, size, and original file structure), and  
+2) **How BTXRD is organized inside this Hugging Face repo** (`saiteja33/MedThink-Seg`, under the `BTXRD/` directory).
+
+---
+
+## 1) Original BTXRD dataset (source)
+
+**Primary reference (Data Descriptor paper):**  
+*“A Radiograph Dataset for the Classification, Localization, and Segmentation of Primary Bone Tumors”* (Scientific Data, 2025).  
+DOI: `10.1038/s41597-024-04311-y` (PDF)
+
+**Original dataset DOI (Figshare):** `10.6084/m9.figshare.27865398`
+
+### What images are in BTXRD?
+- **Modality:** bone **X-ray radiographs** (**JPEG**)  
+- **Body regions covered:** upper limbs, lower limbs, and pelvis  
+- **Views/angles:** frontal, lateral, and oblique  
+- **Task coverage:** segmentation (tumor masks)
+
+### Dataset size and counts
+- **Total size (original ZIP on Figshare):** **801 MB**
+- **Total images:** **3,746**
+  - **Normal:** 1,879
+  - **Tumor:** 1,867  
+    - **Benign:** 1,525  
+    - **Malignant:** 342
+
+### Labels and annotations
+BTXRD provides:
+- **Local annotations** for each *tumor* image:
+  - **bounding boxes** for localization
+  - **polygon masks** for segmentation  in masks dir
+
+**Subtypes (as represented in the paper’s example row):**
+- osteochondroma
+- multiple osteochondromas
+- simple bone cyst
+- giant cell tumor
+- osteofibroma
+- synovial osteochondroma
+- osteosarcoma
+- other bt (other benign tumors)
+- other mt (other malignant tumors)
+
+### BTXRD structure
+The BTXRD download is a **ZIP** that contains:
+```
+BTXRD.zip
+├─ images/              # all radiographs
+│  ├─ IMG000000.jpeg
+│  ├─ IMG000001.jpeg
+│  └─ ...
+├─ masks/         # masks tumor images (segmentation + bbox)
+│  ├─ IMG000000.png
+│  ├─ IMG000001.png
+│  └─ ...
+└─ normal               # images without any tumor
+   ├─ images/              # all radiographs
+   │  ├─ IMG000000.jpeg
+   │  ├─ IMG000001.jpeg
+   │  └─ ...
+   ├─ masks/         # masks tumor images (segmentation + bbox)
+   │  ├─ IMG000000.png
+   │  ├─ IMG000001.png
+   │  └─ ...
+```
+
+**Naming convention:** images start with `IMG`, followed by a **zero‑padded six‑digit** id, and end with `.jpeg`.
+
+---
+
+## How to download / use (recommended)
+
+### A) Download the raw files for BTXRD from this HF repo
+```python
+from huggingface_hub import snapshot_download
+
+local_dir = snapshot_download(
+    repo_id="saiteja33/MedThink-Seg",
+    repo_type="dataset",
+    allow_patterns=["BTXRD/**"],   # only pull BTXRD subdir
+)
+print(local_dir)
+```
+
+### B) Load with 🤗 Datasets (try these options)
+Depending on how the dataset was uploaded, one of these will work:
+
+**Option 1 — direct load from the repo (if a dataset script / supported format is present):**
+```python
+from datasets import load_dataset
+ds = load_dataset("saiteja33/MedThink-Seg", data_dir="BTXRD")
+print(ds)
+```
+
+**Option 2 — treat it as an image-folder dataset (if files are organized as train/validation/test directories):**
+```python
+from datasets import load_dataset
+ds = load_dataset("imagefolder", data_dir="BTXRD")
+print(ds)
+```
+---
+
+## Citation
+If you use BTXRD, cite the Scientific Data paper:
+
+- Shunhan Yao et al. *A Radiograph Dataset for the Classification, Localization, and Segmentation of Primary Bone Tumors.* Scientific Data (2025). DOI: `10.1038/s41597-024-04311-y`
+
+And/or the Figshare dataset record:
+- BTXRD (Figshare). DOI: `10.6084/m9.figshare.27865398`