Skip to content

Latest commit

 

History

History
168 lines (120 loc) · 6.64 KB

SARAS-ESAD.md

File metadata and controls

168 lines (120 loc) · 6.64 KB
Raw

SARAS-ESAD

Dataset Information

The SARAS-ESAD dataset is a dataset for robotic-assisted endoscopic surgical action recognition. It was a challenge in MIDL 2020. The dataset contains a total of 33,398 cases, with its training set providing 22,601 annotated frames, encompassing 28,055 action instances, and the validation set contains 4,574 frames with 7,133 action instances. The test data includes 6,223 annotated frames and 11,565 action instances.

The SARAS-ESAD data presents a challenge for action recognition and analysis in robotic-assisted surgeries, aiming to advance the automation and assistance technologies in robotic surgeries. With detailed action annotations and a variety of surgical scenarios, this dataset not only helps researchers develop and test new algorithms but also deepens our understanding of human-machine interaction patterns during complex surgical procedures. Moreover, it provides rich training material for machine learning models, aiding in improving the models' generalizability and operational precision. The ultimate goal is to enhance surgical efficiency and patient safety through technological advancements.

Dataset Meta Information

Dimensions Modality Task Type Anatomical Structure Anatomical Area Number of Categories Data Volume File Format
2D Endoscopy Action Recognition Intestine Abdominal Cavity 21 33398 jpg, txt

Resolution Details

Dataset Statistics size
min 1920*1080
median 1920*1080
max 1920*1080

Label Information Statistics

Serial Number Categories Number Proportion
0 CuttingMesocolon 555 1.87%
1 PullingVasDeferens 551 1.86%
2 ClippingVasDeferens 90 0.30%
3 CuttingVasDeferens 87 0.29%
4 ClippingTissue 101 0.34%
5 PullingSeminalVesicle 2213 7.48%
6 ClippingSeminalVesicle 124 0.42%
7 CuttingSeminalVesicle 1838 6.21%
8 SuckingBlood 3886 13.13%
9 SuckingSmoke 1092 3.69%
10 PullingTissue 5408 18.27%
11 CuttingTissue 4832 16.32%
12 BaggingProstate 54 0.18%
13 BladderNeckDissection 1627 5.50%
14 BladderAnastomosis 3766 12.72%
15 PullingProstate 895 3.02%
16 ClippingBladderNeck 117 0.40%
17 CuttingThread 100 0.34%
18 UrethraDissection 642 2.17%
19 CuttingProstate 1005 3.40%
20 PullingBladderNeck 618 2.09%

Visualization

File Structure

The dataset file structure is as follows. The Dataset directory contains test, training, and validation data along with labels, located in the test, train, val subdirectories, and test_labels, which include image files and corresponding label files.

Dataset
│
├── test
│   ├── RARP3_frame_10.jpg
│   ├── RARP3_frame_100.jpg
│   ...
│   │
├── test_labels
│   ├── RARP3_frame_10.txt
│   ├── RARP3_frame_100.txt
│   ...
│   │
├── train
│   ├── obj.names
│   ├── set2
│   │   ├── RARP4_frame_1.jpg
│   │   ├── RARP4_frame_1.txt
│   │   ...
│   │   │
│   │
├── val
│   ├── obj
│   │   ├── RARP1_frame_1.jpg
│   │   ├── RARP1_frame_1.txt
│   │   ...
│   │   │
│   ├── obj.names
│   │

Authors and Institutions

Vivek Singh Bawa (Oxford Brookes University, Visual Artificial Intelligence Lab)

Gurkirt Singh (ETH Zurich, Computer Vision Lab)

Francis Kaping'A (Oxford Brookes University, Visual Artificial Intelligence Lab)

Inna Skarga-Bandurova (Oxford Brookes University, Visual Artificial Intelligence Lab)

Elettra Oleari (San Raffaele Hospital, Milan, Italy)

Alice Leporini (San Raffaele Hospital, Milan, Italy)

Carmela Lando (San Raffaele Hospital, Milan, Italy)

Pengfei Zhao (Harbin Institute of Technology)

Xi Xiang (Harbin Institute of Technology)

Gongning Luo (Harbin Institute of Technology)

Kuanquan Wang (Harbin Institute of Technology)

Liangzhi Li (Osaka University)

Bowen Wang (Osaka University)

Shang Zhao (George Washington University)

Li Li (George Washington University)

Armando Stabile (San Raffaele Hospital, Milan, Italy)

Francesco Setti (University of Verona)

Riccardo Muradore (University of Verona)

Fabio Cuzzolin (Oxford Brookes University, Visual Artificial Intelligence Lab)

Source Information

Official Website: https://saras-esad.grand-challenge.org

Download Link: https://saras-esad.grand-challenge.org

Article Address: https://arxiv.org/abs/2104.03178

Publication Date: 2020-01

Citation

@misc{bawa2021saras,
      title={The SARAS Endoscopic Surgeon Action Detection (ESAD) dataset: Challenges and methods}, 
      author={Vivek Singh Bawa and Gurkirt Singh and Francis KapingA and Inna Skarga-Bandurova and Elettra Oleari and Alice Leporini and Carmela Landolfo and Pengfei Zhao and Xi Xiang and Gongning Luo and Kuanquan Wang and Liangzhi Li and Bowen Wang and Shang Zhao and Li Li and Armando Stabile and Francesco Setti and Riccardo Muradore and Fabio Cuzzolin},
      year={2021},
      eprint={2104.03178},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

Original introduction article is here.