Correlation Networks of Spinal Motor Neurons that Innervate Lower Limb Muscles during a Multi-Joint Isometric Task

This repository contains all the raw HD-EMG data, the manually edited motor unit spike trains and the data selected to build networks of common inputs to motor neurons.<br><br>Contact:<br>Francois Hug ([e-mail])Simon Avrillon ([e-mail])<br><br>=================================<br><br>FILE NAME: Raw_data.zip <br><br>This compressed file contains the raw data together with the motor unit spike trains manually edited.<br><br>The HDsEMG signals were decomposed into motor unit spike trains using convolutive blind-source separation as described in the manuscript (custom made matlab script). After the automatic identification of the motor units, the spike trains were visually checked for false positives and false negatives using the DEMUSE tool (The University of Maribor, Slovenia).<br><br><br> Filename: [Participant number]_135_175.otb+_[muscle]_Niter150_FastICA_edited_final.mat<br><br>Participant: from #1 to #10.<br>Muscle: GM (gastrocnemius medialis), GL (gastrocnemius lateralis), VL (vastus lateralis), VM (vastus medialis), BF (biceps femoris), ST (semitendinosus)<br><br>These files can be opened with Matlab (Mathworks) or the DEMUSE tool. VARIABLES (only the variables used in the manuscript are described below) :<br><br> - DecompRuns: number of decomposition runs - Note that this information is incorrect on these files<br> - discardChannelsVec: discarded channels for the 13*5 EMG grid (ELSCH064NM2, SpesMedica, Battipaglia). "1" means that the channnel has been discarded due to noise/artifacts. Note that one electrode was absent on a corner [1,1]. <br>- fsamp: sampling rate<br> - IED: inter-electrode distance <br>- IPTs: Innervation Pulse Train (IPT, i.e. train of motor unit discharge times as estimated by the decomposition technique) <br>- MUPulses (1*number of motor units array). For each identified units, there is a vector of motor units discharge times (in datapoint).<br> - PNR: Pulse-to-noise ratio for each identified MU - Of note, these values are calculated for the whole signal and therefore they do not correspond to the PNR of the processed 10-s windows.<br>- ref_signal: force signal (only z component) <br>- SIG (13*5 cell array): raw EMG signal organised on a 13*5 cell array. Each cell corresponds to an EMG channel.<br> - SIGlength: duration of the signal in seconds<br>=================================<br><br>FILE NAME: Processed_data.zip <br><br>This compressed file contains the processed data for the two different analyses, i.e. repeatability and construction of the networks.<br>Data for the reliability between matrices of correlation:<br>Name: ’S ‘number of participant’ _repro.mat<br><br>Structure:data.muscle: name of the motor unit with the name of the muscle and the number of the motor unit. e.g., BF1.<br>data.firings: matrix of motor unit’s discharge times. This is a binary file where each discharge time = 1.<br>data.time: time for the motor unit’s discharge times.<br>data.Matcorrel_Win1: matrix of correlations for each pair of motor units over the first window.<br>data.Matcorrel_Win2: matrix of correlations between each pair of motors unit over the second window.<br><br>Data for the network analysis:<br>Name: ’S ‘number of participant’ _data.mat<br><br>Structure:<br>data.Muscle: name of the motor unit with the name of the muscle and the number of the motor unit. e.g., BF1.<br>data.firings: matrix of motor unit’s discharge times. This is a binary file where each discharge time = 1.<br>data.time: time for the motor unit’s discharge times.<br>data.matcorrel: matrix of correlations for each pair of motor units.<br>data.fsamp: sample frequency, I.e., 2048 Hz.<br><br><br>

本仓库包含全部原始高密度肌电图（HD-EMG）数据、经人工校正的运动单位锋电位序列（motor unit spike trains），以及用于构建运动神经元共同输入网络的筛选数据集。

联系方式：
Francois Hug（[e-mail]）、Simon Avrillon（[e-mail]）

=================================

文件名：Raw_data.zip

该压缩包包含原始数据与经人工校正的运动单位锋电位序列。

研究人员采用卷积盲源分离技术（详见论文），将高密度肌电信号分解为运动单位锋电位序列，所用代码为定制化MATLAB脚本。在自动识别运动单位后，通过DEMUSE工具（斯洛文尼亚马里博尔大学开发）对锋电位序列进行人工核查，以剔除假阳性与假阴性结果。

文件命名规则：[受试者编号]_135_175.otb+_[肌肉名称]_Niter150_FastICA_edited_final.mat

受试者范围：编号1至10。
肌肉类型：腓肠肌内侧头（GM, gastrocnemius medialis）、腓肠肌外侧头（GL, gastrocnemius lateralis）、股外侧肌（VL, vastus lateralis）、股内侧肌（VM, vastus medialis）、股二头肌（BF, biceps femoris）、半腱肌（ST, semitendinosus）

此类文件可通过MATLAB或DEMUSE工具打开。下文仅介绍论文中用到的核心变量：

- DecompRuns：分解运行次数——请注意，该文件中此信息存在错误
- discardChannelsVec：13×5肌电阵列（ELSCH064NM2，SpesMedica，Battipaglia）的丢弃通道标识，“1”代表该通道因噪声或伪影被丢弃。需注意，电极阵列一角的[1,1]位置缺失一枚电极。
- fsamp：采样率
- IED：电极间距
- IPTs：神经冲动脉冲序列（Innervation Pulse Train, IPT），即通过分解技术估算得到的运动单位放电时序序列
- MUPulses（1×运动单位数量数组）：针对每个识别出的运动单位，存储其放电时间（以数据点为单位）的向量
- PNR：各识别运动单位的脉冲信噪比——需注意，该值为全信号计算所得，并非针对处理后的10秒窗口的脉冲信噪比
- ref_signal：测力信号（仅z轴分量）
- SIG（13×5元胞数组）：按13×5阵列组织的原始肌电信号，每个元胞对应一个肌电通道
- SIGlength：信号时长（单位：秒）

=================================

文件名：Processed_data.zip

该压缩包包含两类分析所用的处理后数据，即可重复性分析与网络构建分析。

相关性矩阵可靠性分析所用数据：
命名格式：'S_[受试者编号]_repro.mat'

数据结构：
data.muscle：运动单位命名，格式为「肌肉名称+运动单位编号」，例如BF1
data.firings：运动单位放电时间矩阵，为二进制文件，每个放电时刻对应值为1
data.time：运动单位放电时刻对应的时间轴
data.Matcorrel_Win1：第一个窗口内每对运动单位间的相关性矩阵
data.Matcorrel_Win2：第二个窗口内每对运动单位间的相关性矩阵

网络分析所用数据：
命名格式：'S_[受试者编号]_data.mat'

数据结构：
data.Muscle：运动单位命名，格式为「肌肉名称+运动单位编号」，例如BF1
data.firings：运动单位放电时间矩阵，为二进制文件，每个放电时刻对应值为1
data.time：运动单位放电时刻对应的时间轴
data.matcorrel：每对运动单位间的相关性矩阵
data.fsamp：采样频率，即2048 Hz