Datasets

Download K2MUSE

  The complete dataset and related code have been uploaded to Kaggle. Please click the following link to download and use them.

  Kaggle: K2MUSE: A Human Lower-Limb Multimodal Dataset

  ‍

Data Organization

  The raw data and processed data are stored in the ‘SourceData’ and ‘ProcessedData’ folders, respectively. The specific structure of the dataset is as follows:

DatasetStructure(full)

Sourece Data

  All raw, unprocessed data are exported and stored in the ‘SourceData’ folder. The subfolders within SourceData - Vicon, Noraxon, and ELONXI - correspond to data collected by different devices.

  The folder structure is as follows:

DatasetStructure(Source)

Vicon Folder: Participant/Condition/Participant+Condition+AmbulationMode+Angle+Speed+TrialNumber.c3d
Note: This folder stores kinematic and kinetic data of the lower limbs.
(Participant)=One of thirty able - bodied individuals (P*)
(Condition)=Different acquisition conditions, including ideal condition (Ideal), muscle fatigue (Fatigue), electrode shifts (Shift), and inter - day variations (InterDay). The Ideal condition serves as a reference compared to the other non - ideal conditions.
(AmbulationMode)=Walking on different terrains, including level - ground walking (LG), ascending ramp walking (AS), and descending ramp walking (DS).
(Angle)=The incline angles of the treadmill include , , and 10°.
(Speed)=Walking speeds include 0.5 m/s, 1.0 m/s, and 1.5 m/s.
(Trial Number)=The number of walking trials: For the muscle fatigue experiment, it ranges from 1 to 10, while for the other conditions, the trials range from 1 to 5.
(.c3d)=The data format exported by Nexus software, used for storing kinematic and kinetic data.

  ‍

ELONXI Folder: Participant/Condition/Ambulation Mode/US_Channel_Speed_TrialNumber.txt
Note: This folder stores data of the left leg collected using the AUS device (ELONXI).
(Participant)=One of thirty able - bodied individuals (P*)
(Condition)=Different acquisition conditions include ideal condition (Ideal), muscle fatigue (Fatigue), electrode shifts (ShiftInitial, ShiftUpward, ShiftDownward, ShiftLeftward, ShiftRightward), and inter - day differences (InterDay)
(AmbulationMode)=Different ramps and angles, represented as "Terrain_Angle," including AS_5, AS_10, DS_5, DS_10, and LG_0
(US)=The abbreviation for Ultrasound
(Channel)=The device channel numbers, including: CH1, CH2, CH3, and CH4
(Speed)=Walking speeds: S_0x5 corresponds to 0.5 m/s, S_1x0 corresponds to 1.0 m/s, and S_1x5 corresponds to 1.5 m/s
(TrialNumber)=The number of walking trials: For muscle fatigue experiments, it ranges from 1 to 10; for other conditions, it ranges from 1 to 5
(.txt)=The data format saved by the ELONXI device, used for storing AUS data
Noraxon Folder: Participant/Condition - AmbulationMode - Speed - TrialNumber.mat
Note: This folder stores data of both legs collected using the sEMG device (Noraxon Ultium).
(Participant)=One of thirty able - bodied individuals (P*)
(Condition)=Different acquisition conditions include ideal condition (Ideal), muscle fatigue (Fatigue), electrode shifts (ShiftInitial, ShiftUpward, ShiftDownward, ShiftLeftward, ShiftRightward), and inter - day differences (InterDay)
(AmbulationMode)=Different ramps and angles, represented as "Terrain - Angle," including AS - 5, AS - 10, DS - 5, DS - 10, and LG - 0
(Speed)=Walking speeds: 0x5 corresponds to 0.5 m/s, 1x0 corresponds to 1.0 m/s, and 1x5 corresponds to 1.5 m/s
(TrialNumber)=The number of walking trials: For muscle fatigue experiments, it ranges from 1 to 10; for other conditions, it ranges from 1 to 5
(.mat)=The data format exported by MR software, used for storing sEMG data

Processed Data

  The experimental data is saved in uniform MATLAB structures, with each subject corresponding to a separate “P*.mat” file. Each “P*.mat” file contains two identical structured structs: “OriginalData” and “NormalizedData,” which correspond to the raw extracted data and the gait-normalized data, respectively. The folder structure is as follows:

DatasetStructure(Processed)

  ‍

OriginalData.(Condition).(AmbulationMode).(Speed).(TrialNumber).(Datatype).(Variable)
(Condition)=Different acquisition conditions, including ideal condition (Ideal), muscle fatigue (Fatigue), electrode shifts (ShiftInitial, ShiftUpward, ShiftDownward, ShiftLeftward, ShiftRightward), and inter - day variations (InterDay)
(AmbulationMode)=Different ramps and angles, represented as "Terrain - Angle," including AS_5, AS_10, DS_5, DS_10, and LG_0
(Speed)=Walking speeds: S_0x5 corresponds to 0.5 m/s, S_1x0 corresponds to 1.0 m/s, and S_1x5 corresponds to 1.5 m/s
(TrialNumber)=The number of walking trials: For the muscle fatigue experiments, it ranges from 1 to 10, while for the other conditions, the trials range from 1 to 5
(Datatype)=Markers: Position of markers in global coordinates (unit: m).
Angles: Tracks the joint angle in three dimensions (unit: deg).
Forces: Tracks the resultant force at each joint (unit: N/kg).
Moments: Tracks the moment about each joint (unit: N*m/kg). Full documentation Link for above datatype
Powers: The power output by the joint (unit: W/kg).
Events: Bilateral gait events, including heel strike and foot off, categorized by frame and time.
USData: The 4 - channel AUS data of the left leg, with dimensions of N*1, representing N frames of signals.
EMGData: Bilateral sEMG data, with dimensions of M*13, representing 13 channels, each with M sampling points.
(Variable)=The final field for all data types.
For gait events, EventTime represents the gait events based on time, while EventFrame represents the gait events based on frames.
For marker trajectories, kinematic, and kinetic data, each row represents a sampling point, and each column represents a data dimension.
For AUS data, the dimension of each frame is 4*1000, with the first to fourth rows corresponding to channels 1 to 4. For EMG data, columns 1 to 13 correspond to channels 1 to 13. Therefore, channels 1 to 9 correspond to the right leg, and channels 10 to 13 correspond to the left leg.

  ‍

NormalizedData.(Condition).(AmbulationMode).(Speed).(TrialNumber).(Datatype).(Variable)
(Condition)=Different acquisition conditions, including ideal condition (Ideal), muscle fatigue (Fatigue), electrode shifts (ShiftInitial, ShiftUpward, ShiftDownward, ShiftLeftward, ShiftRightward), and inter-day variations (InterDay).
(AmbulationMode)=Different ramps and angles, represented as "Terrain-Angle," including AS_5, AS_10, DS_5, DS_10, and LG_0.
(Speed)=Walking speeds: S_0x5, S_1x0, and S_1x5 correspond to speeds of 0.5 m/s, 1.0 m/s, and 1.5 m/s, respectively.
(TrialNumber)=The number of walking trials: For the muscle fatigue experiments, it ranges from 1 to 10, while for the other conditions, the trials range from 1 to 5.
(Datatype)=Markers: Position of markers in global coordinates (unit: m).
Angles: Tracks the joint angle in three dimensions (unit: deg).
Forces: Tracks the resultant force at each joint (unit: N/kg).
Moments: Tracks the moment (or Torque) about each joint (unit: N*m/kg).
Powers: The power output by the joint (unit: W/kg).
Events: Bilateral heel strikes after stable walking, based on frames.
USData: AUS data segmented by left leg heel strikes, with dimensions of N*1, representing N gait cycles.
EMGData: sEMG data segmented by bilateral heel strikes, divided into "Right" and "Left"
(Variable)=The final field for all data types.
For heel strikes, the variables include LStartFrame, LEndFrame, RStartFrame, and REndFrame, with dimensions of 1*M, representing M gait events.
For marker trajectories, kinematic, and kinetic data, each variable includes: rsGaitData - All resampled gait data for the current trial; MeanValues - The average of all gait data for that trial; StdValues - The root mean square error for all gait data for that trial.
For AUS data, each gait corresponds to M*1 data, where M represents the number of frames for that gait. For EMG data, each gait in "Right" or "Left" corresponds to P*9 or P*4 data, where P represents the number of sampling points for that gait.

Figure Data

  The “figData.mat” file in the folder is used by the “scriptPlot.mlx” script to plot the average joint angles and joint moments for all participants.

Pictures of Participants

Ideal Conditions & Muscle Fatigue

  The following are the frontal views of 30 subjects in the ideal conditions and muscle fatigue experiments:

ParticipantsPictures

Electrode Shifts

  The following are the frontal views of 8 participants in the electrode shifts experiment. From left to right, each column corresponds to the following shift directions: ShiftInitial, ShiftUpward, ShiftDownward, ShiftLeftward, and ShiftRightward.

ParticipantsPictures_Shift

  ‍

Videos of Experiments

  The following are videos of participants completing the ideal conditions and muscle fatigue experiments.

  Note: The experimental protocol for inter-day variations is the same as for the ideal condition, so no photos or videos were recorded for this condition.


  ‍