Difference between revisions of "Gait events"

Revision as of 13:42, 31 May 2016

Gait Event Detection in Real-World Environments for Long-Term Applications

This is a placeholder for an automated service that will provide temporal estimation of gait events from accelerometer signals. The service would take accelerometer data (in the specified format) as input and provide the estimated events as output.

Until this service is implemented, please send your data in the specified format to siddhartha.khandelwal@hh.se and nicholas.wickstrom@hh.se.

Specifications

Activities: Walking and running

Placement of 3-axis Accelerometer: Anywhere around the ankle in any orientation.

Sensitivity of the Accelerometer: (+-) 4g or more. Please specify. Please also check if the accelerometer signal has saturated during intense activity such as running.

Sampling Frequency: Preferred - 128 Hz [A Sampling frequency of 50Hz and above is acceptable. Please specify.]

Input data: AccX, AccY, AccZ [Accelerometer signals from the X, Y and Z axes, respectively.]

Input data format: abc.mat or abc.csv [Data in Matlab format or comma separated text file.]

NOTE: The data should consist of ONLY walking and running segments of the signal. Segments corresponding to inactivity or any other activity should be removed from the signals prior to data submission.

[1] Siddhartha Khandelwal; Nicholas Wickström, "Gait Event Detection in Real-World Environment for Long-Term Applications: Incorporating Domain Knowledge into Time-Frequency Analysis," in IEEE Transactions on Neural Systems and Rehabilitation Engineering , vol.PP, no.99, pp.1-1

IEEE url: [1] DiVA url: [2]

[2] Siddhartha Khandelwal; Nicholas Wickström, "Identification of Gait Events using Expert Knowledge and Continuous Wavelet Transform Analysis", in BIOSIGNALS 2014, 7th International Conference on Bio-inspired Systems and Signal Processing, Angers, France, March 3-6, 2014

DiVA url: [3]

MAREA (Movement Analysis in Real-world Environments using Accelerometers) Gait Database

"Figure 1: Position and orientation of each accelerometer at the beginning of each experiment. The accelerometer at the right ankle is attached arbitrarily without any pre-defined orientation. The FSRs are instrumented into the shoe soles to collect the ground truth. The data from all sensors is sampled at a frequency of 128 Hz."

The MAREA gait database comprises of gait activities in different real-world environments as shown in the table below. 20 healthy adults (12 males and 8 females, average age: 33.4 +- 7 years, average weight: 73.2 +- 10.9 kg, average height: 172.6 +- 9.5 cm) participated in the study that was approved by the Ethical Review Board of Lund, Sweden. Each subject had a 3-axes Shimmer3 (Shimmer Research, Dublin, Ireland) accelerometer (+- 8g) attached to their waist, left wrist and left and right ankles using elastic bands and velcro straps. Figure 1 shows the position and orientation of each accelerometer at the beginning of each experiment. The X and Y axes of the accelerometer positioned on the waist and the Y and Z axes of the accelerometers positioned on the left wrist and left ankle were aligned with the sagittal and transverse planes, respectively. The accelerometer on the right ankle was casually attached without any predefined alignment to simulate a daily life scenario. The subjects were provided shoes that were instrumented with piezo-electric force sensitive resistors (FSRs), fixed at the extreme ends of the sole in order to provide the ground truth values for HS and TO. The data from accelerometer and FSRs was sampled at a frequency of 128Hz, and the FSR output was stored locally on the Shimmer3 microSD card using an external expansion board. The accelerometer signals obtained from different body locations were synchronized manually. Timings of the switch from walking to running were noted down during the experiments, in order to segregate the dataset into walking and running segments.

of 0.4km/hr every minute

There are two matrices provided, namely, Indoor Experiment timings and Outdoor Experiment timings which contain the timing information of the activities and are included in a folder called Activity Timings.zip (download link below):

"Example of how to use the sample numbers given in Indoor Experiment Timings matrix to extract desired activity data. This figure shows the entire Indoor data collected from accelerometer positioned at Right Ankle for Subject 5. "

• Indoor Experiment Timings: This 11(rows) x 8(columns) matrix consists of sample numbers corresponding to the start and end of an activity, for a given subject, for the Indoor Experiments, i.e.
  • Treadmill (flat)
  • Treadmill (slope)
  • Indoor flat space.

The figure below explains how to extract the sample nos. corresponding to an activity for a given subject, from the Indoor Experiment timings matrix.

• Outdoor Experiment Timings: This 9(rows) x 3(columns) matrix consists of sample numbers corresponding to the start and end of an activity, for a given subject, for the Outdoor Street Experiments. The figure below explains how to extract the sample nos. corresponding to an activity for a given subject, from the Outdoor Experiment timings matrix.

"Figure 2: Example of the accelerometer data from the X, Y and Z axis for Sub5_RF"

The Subject data files are provided in two formats, namely, .mat format (Subject Data_mat format.zip) and .txt format Subject Data_txt format).zip. The naming convention of the files is: Sub<number>_<position>, where:

• <number>: stands for Subject number and ranges from 1 to 20.
• <position>: stands for the position of the accelerometer on the body as shown in Figure 1. The positions are:
  • LF - Left Ankle
  • RF - Right Ankle
  • Wrist
  • Waist

For each Subject file, eg. Sub5_RF, the accelerometer data from the 3 axis accelerometer is stored in 3 columns (separated using comma in the .txt files), each named as:

• accX - data from X - axis
• accY - data from Y - axis
• accZ - data from Z - axis

The Ground Truth is provided as a 11x7 structure that gives the timing of the Heel-Strike and Toe-Off events extracted from the FSR signals. This timing information is provided in terms of sample numbers and is relative to each activity. The structure consists of 7 fields that have already been segregated using the Indoor Experiment Timings and Outdoor Experiment Timings explained above:

• treadWalk - treadmill (flat) walk

• treadIncline - treadmill (slope) walk

• treadWalknRun - treadmill (flat) walk & run

• indoorWalk - indoor (flat space) walk

• indoorWalknRun - indoor (flat space) walk & run

• outdoorWalk - outdoor street walk

• outdoorWalknRun - outdoor street walk & run

Each row of the structure represents a Subject. For Indoor activities, Row 1 -> Subject1, ..., Row 11 -> Subject 11 and so on. For outdoor activities, Row 1 -> Subject 12,..., Row 9 -> Subject 20. As an example: GroundTruth(1).treadWalk provides the ground truth data for Subject 1 for treadmill (flat) walk. This is a structure with fields:

• SubIdx - Subject number. Ranges from Sub1 - Sub20

• LF_HS - Sample numbers of Heel-Strike event from Left Foot FSR signal.

• LF_TO - Sample numbers of Toe-Off event from Left Foot FSR signal.

• RF_HS - Sample numbers of Heel-Strike event from Right Foot FSR signal.

• RF_TO - Sample numbers of Toe-Off event from Right Foot FSR signal.

• Link to download timing information of the activities (both .txt and .mat format): File:Activity Timings.zip

• Link to download the Subject data files (.mat format): [To be made public very soon]

• Link to download the Subject data files (.txt format): [To be made public very soon]

[1] Siddhartha Khandelwal; Nicholas Wickström, "Evaluation of the performance of accelerometer-based gait event detection algorithms in different real-world scenarios using the MAREA gait database," submitted to Gait & Posture, 2016.