Subglottal Pressure Experiments with Ex-vivo Porcine Larynges

The dataset consists of 288 high-speed video endoscopy (HSV) porcine larynx recordings. Phonation was recorded at differing subglottal pressure levels. Each video was segmented using the software Glottis Analysis Tools, and trajectories were computed from each segmentation.

The following files are included:

videos.zip - larynx recordings in .mp4-format

measures.csv - table consisting metadata

larynx: numeric identifier of the larynx

gap: Length of the inserted metal shim (mm)

glottis_length: length of the glottal midline (px)

psub: measured subglottal pressure (Pa)

adduction_left, adduction_right: arytenoid torques (mNm)

trajectories.npy - trajectories (longitudinal 25%, 50%, 75%) as numpy array

0: left - posterior

1: left - medial

2: left - anterior

3: right - posterior

4: right - medial

5: right - anterior

Usage:

# load trajectoriesimport numpy as nparr = np.load('trajectories.npy')

# load meta-dataimport pandas as pddf = pd.read_csv('measures.csv')

本数据集包含288段高速视频内窥镜（high-speed video endoscopy, HSV）拍摄的猪喉部录制片段。录制过程中，喉部发声时的声门下压水平各不相同。所有视频均通过声门分析工具（Glottis Analysis Tools）进行分割，并从每个分割结果中计算得到运动轨迹。

本次发布包含以下文件：
- videos.zip：.mp4格式的喉部录制视频
- measures.csv：包含元数据的数据表，各字段说明如下：
- larynx：喉部的数字标识符
- gap：插入的金属垫片长度（单位：毫米，mm）
- glottis_length：声门中线长度（单位：像素，px）
- psub：实测声门下压（单位：帕斯卡，Pa）
- adduction_left、adduction_right：分别为左侧、右侧杓状软骨扭矩（单位：毫牛米，mNm）
- trajectories.npy：以NumPy数组存储的运动轨迹（对应纵向25%、50%、75%位置），数组各索引对应的部位如下：
0: 左侧-后部
1: 左侧-中部
2: 左侧-前部
3: 右侧-后部
4: 右侧-中部
5: 右侧-前部

使用示例：
python
# 加载运动轨迹
import numpy as np
arr = np.load('trajectories.npy')

# 加载元数据
import pandas as pd
df = pd.read_csv('measures.csv')