---
license: mit
pretty_name: Open Knowledge-Based Planning
dataset_info:
  features:
  - name: ct
    dtype:
      array3_d:
        shape:
        - 128
        - 128
        - 128
        dtype: float32
  - name: dose
    dtype:
      array3_d:
        shape:
        - 128
        - 128
        - 128
        dtype: float32
  - name: voxel_dimensions
    sequence: float32
    length: 3
  - name: patient
    dtype: string
  - name: possible_dose_mask
    dtype:
      array3_d:
        shape:
        - 128
        - 128
        - 128
        dtype: float32
  - name: structure_masks
    dtype:
      array4_d:
        shape:
        - 128
        - 128
        - 128
        - 10
        dtype: float32
  - name: structure_mask_names
    sequence: string
    length: 10
  splits:
  - name: train
    num_bytes: 23540972292
    num_examples: 200
  - name: validation
    num_bytes: 4708194480
    num_examples: 40
  - name: test
    num_bytes: 11770486200
    num_examples: 100
  download_size: 363683542
  dataset_size: 40019652972
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
  - split: test
    path: data/test-*
---

# Open Knowledge-Based Planning
🤗 mirror of the Open Knowledge-Based Planning (OpenKBP) dataset, taken from [GitHub](https://github.com/ababier/open-kbp)