Skip to content

Latest commit

 

History

History
94 lines (67 loc) · 4.06 KB

File metadata and controls

94 lines (67 loc) · 4.06 KB

Ultrasound Nerve Segmentation

Dataset Information

The dataset originates from the Kaggle Ultrasound Nerve Segmentation Challenge, which includes 11,143 ultrasound images with a resolution of 580x420 pixels, manually annotated by clinical experts to create mask images. Of these samples, 5,635 images were used for training and 5,508 images for testing, with the training set labeled and the test set unlabeled. Even the bravest patients can feel apprehensive at the mention of surgery. Surgical procedures inevitably bring discomfort and often involve significant postoperative pain. Currently, patient pain is typically managed using anesthetics, which come with many unwanted side effects. The sponsors of the Kaggle Ultrasound Nerve Segmentation Challenge are working to improve pain management by using indwelling catheters that block or mitigate pain at its source. Pain management catheters reduce reliance on anesthetics and accelerate patient recovery. Accurately identifying nerve structures in ultrasound images is a crucial step for the effective insertion of pain management catheters in patients. In this competition, Kaggle participants are challenged to build a model that can identify nerve structures in a dataset of neck ultrasound images. Doing so can improve the placement of catheters, contributing to a more pain-free future.

Dataset Meta Information

Dimensions Modality Task Type Anatomical Structures Anatomical Area Number of Categories Data Volume File Format
2D Ultra Sound Segmentation Cervical nerves Head and Neck 1 5635 with annotation, 5508 without annotation. tif

Resolution Details

Dataset Statistics size
min (580, 420)
median (580, 420)
max (580, 420)

Label Information Statistics

Segmentation Class Cervical nerves
Case Count 2323
Detection Rate 41%
Min Volume (cm³) 2684
Median Volume (cm³) 6976
Max Volume (cm³) 17439

Note: Label statistics are based on the 5635 labeled images in the training set.

Visualization

Left: Ultrasound image; Right: Segmentation annotation.

File Structure

The file structure of the Ultrasound-Nerve-Segmentation dataset is as follows. There are two folders: "train" for the training set and "test" for the test set.

ultrasound-nerve-segmentation
├── sample_submission.csv
├── test
│   ├── 9.tif
│   ├── 99.tif
│   ├── 999.tif
│   └── ...
└── train
    ├── 33_77_mask.tif
    ├── 33_77.tif
    ├── 33_78_mask.tif
    ├── 33_78.tif
    ├── 33_79.tif
    └── ...

Authors and Institutions

kaggle

Source Information

Official Website: https://kaggle.com/competitions/ultrasound-nerve-segmentation

Download Link: https://www.kaggle.com/c/ultrasound-nerve-segmentation/data

Article Address: -

Publication Date: 2016.5.19

Citation

@misc{ultrasound-nerve-segmentation,
    author = {Anna Montoya, Hasnin, kaggle446, shirzad, Will Cukierski, yffud},
    title = {Ultrasound Nerve Segmentation},
    publisher = {Kaggle},
    year = {2016},
    url = {https://kaggle.com/competitions/ultrasound-nerve-segmentation}
}

Original introduction article is here.