metadata

dataset_info:
  features:
    - name: image
      dtype: image
    - name: organ
      dtype: image
    - name: gonogo
      dtype: image
    - name: id
      dtype: string
  splits:
    - name: train
      num_bytes: 197384771
      num_examples: 785
    - name: test
      num_bytes: 58310857
      num_examples: 230
  download_size: 255917924
  dataset_size: 255695628
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*

Dataset Structure

This dataset contains vision data from cholecystectomy surgery (gallbladder removal).

Data Fields

image: The PIL image of the surgery view.
gonogo: The (360,640) label of background (0), safe (1), and unsafe (2).
organs: The (360,640) label of background (0), liver (1), gallbladder (2), and hepatocystic triangle (3).

Data Splits

train: 785 samples (from 92 videos)
test: 230 samples (from 26 videos)
Total: 1015 samples (from 118 videos in total)

Usage

from datasets import load_dataset
train_dataset = load_dataset("BrachioLab/cholec", split="train")
test_dataset = load_dataset("BrachioLab/cholec", split="test")

Data split

To note that we randomly split the data 8:2 so that our train/test splits have the same distribution. This could have overlap with other datasets that use cholec80 and M2CAI2016. Please take the overlap into consideration when you use auxiliary data for training.

Videos in the training set: 'M2CCAI2016_video103', 'cholec80_video44', 'M2CCAI2016_video92', 'cholec80_video47', 'cholec80_video59', 'cholec80_video74', 'M2CCAI2016_video98', 'cholec80_video65', 'M2CCAI2016_video81', 'cholec80_video05', 'M2CCAI2016_video90', 'cholec80_video13', 'M2CCAI2016_video83', 'M2CCAI2016_video115', 'cholec80_video22', 'cholec80_video19', 'M2CCAI2016_video114', 'cholec80_video23', 'M2CCAI2016_video86', 'cholec80_video53', 'cholec80_video39', 'M2CCAI2016_video121', 'cholec80_video51', 'M2CCAI2016_video87', 'cholec80_video08', 'cholec80_video07', 'cholec80_video27', 'cholec80_video12', 'M2CCAI2016_video84', 'M2CCAI2016_video106', 'cholec80_video15', 'cholec80_video61', 'cholec80_video43', 'M2CCAI2016_video117', 'M2CCAI2016_video109', 'cholec80_video46', 'cholec80_video35', 'cholec80_video18', 'cholec80_video37', 'M2CCAI2016_video112', 'M2CCAI2016_video99', 'cholec80_video67', 'cholec80_video71', 'M2CCAI2016_video104', 'cholec80_video50', 'M2CCAI2016_video110', 'M2CCAI2016_video100', 'M2CCAI2016_video102', 'M2CCAI2016_video94', 'cholec80_video80', 'cholec80_video20', 'cholec80_video34', 'M2CCAI2016_video96', 'cholec80_video69', 'cholec80_video25', 'cholec80_video60', 'cholec80_video64', 'cholec80_video48', 'M2CCAI2016_video118', 'M2CCAI2016_video108', 'cholec80_video73', 'M2CCAI2016_video101', 'cholec80_video77', 'cholec80_video79', 'M2CCAI2016_video105', 'cholec80_video54', 'cholec80_video30', 'cholec80_video49', 'cholec80_video14', 'cholec80_video62', 'M2CCAI2016_video120', 'M2CCAI2016_video88', 'cholec80_video42', 'cholec80_video09', 'cholec80_video76', 'M2CCAI2016_video93', 'M2CCAI2016_video91', 'cholec80_video45', 'cholec80_video68', 'M2CCAI2016_video111', 'cholec80_video32', 'cholec80_video70', 'M2CCAI2016_video119', 'cholec80_video41', 'cholec80_video75', 'cholec80_video38', 'M2CCAI2016_video89', 'cholec80_video16', 'cholec80_video26', 'cholec80_video72', 'cholec80_video29', 'cholec80_video21'

Videos in the test set: 'cholec80_video66', 'cholec80_video56', 'cholec80_video17', 'cholec80_video55', 'M2CCAI2016_video113', 'cholec80_video06', 'cholec80_video02', 'cholec80_video78', 'cholec80_video01', 'cholec80_video40', 'cholec80_video04', 'cholec80_video11', 'M2CCAI2016_video116', 'M2CCAI2016_video95', 'cholec80_video33', 'cholec80_video57', 'cholec80_video03', 'cholec80_video28', 'cholec80_video31', 'cholec80_video52', 'cholec80_video24', 'M2CCAI2016_video107', 'cholec80_video63', 'M2CCAI2016_video97', 'cholec80_video36', 'cholec80_video58'

Ciations

For the combined gonogo and organs labels, please cite FIX:

@misc{jin2024fix,
    title={The FIX Benchmark: Extracting Features Interpretable to eXperts},
    author={Helen Jin and Shreya Havaldar and Chaehyeon Kim and Anton Xue and Weiqiu You and Helen Qu and Marco Gatti and Daniel A Hashimoto and Bhuvnesh Jain and Amin Madani and Masao Sako and Lyle Ungar and Eric Wong},
    year={2024},
    eprint={2409.13684},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

Please also cite the original datasets:

Cholec80

@misc{twinanda2016endonetdeeparchitecturerecognition,
      title={EndoNet: A Deep Architecture for Recognition Tasks on Laparoscopic Videos}, 
      author={Andru P. Twinanda and Sherif Shehata and Didier Mutter and Jacques Marescaux and Michel de Mathelin and Nicolas Padoy},
      year={2016},
      eprint={1602.03012},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/1602.03012}, 
}

M2CAI2016

@misc{twinanda2016endonetdeeparchitecturerecognition,
      title={EndoNet: A Deep Architecture for Recognition Tasks on Laparoscopic Videos}, 
      author={Andru P. Twinanda and Sherif Shehata and Didier Mutter and Jacques Marescaux and Michel de Mathelin and Nicolas Padoy},
      year={2016},
      eprint={1602.03012},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/1602.03012}, 
}

@misc{stauder2017tumlapcholedatasetm2cai,
      title={The TUM LapChole dataset for the M2CAI 2016 workflow challenge}, 
      author={Ralf Stauder and Daniel Ostler and Michael Kranzfelder and Sebastian Koller and Hubertus Feußner and Nassir Navab},
      year={2017},
      eprint={1610.09278},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/1610.09278}, 
}