metadata

pretty_name: LIDC-IDRI Text-Annotated
license: cc-by-nc-sa-4.0
language:
  - en
task_categories:
  - image-segmentation
  - image-to-text
multilinguality: monolingual
source_datasets:
  - LIDC-IDRI (TCIA)
size_categories:
  - 10K<n<100K

🩺 LIDC-IDRI Text-Annotated

We release a text-annotated version of the LIDC-IDRI dataset, where each annotation is carefully curated from structured metadata provided by radiologists (e.g., malignancy, size, shape, margin, texture, spiculation, etc.).

This enables new research directions in:

Multi-modal learning (image + text)
Text-guided medical image segmentation

Includes

Radiologist Nodule annotations (radiologist contours, malignancy scores)
Natural language descriptions derived from metadata

This dataset is based on the publicly available LIDC-IDRI release from TCIA. Text annotations are derived from radiologist-provided metadata.