Electromyography Data

Summary 

 Surface EMG (sEMG) signals collected during activities of daily life (ADL) provide better insights toward understanding neuromuscular disorders, persons with limb disabilities, aging adults and neuromotor deficits. Hand movement and control mechanism analysis may improve the design of prosthetic devices, realistic biomechanical hands, and rehabilitation therapy. Towards this goal, we present a sEMG signal database corresponding to the Indian population named “ElectroMyography Analysis of Hand Activities - DataBase -1 (EMAHA-DB1).”  The dataset can be used for classification studies and statistical analysis of sEMG signals.

Description: 

Study participants: 

The institutional ethics committee of Indian Institute of Information Technology Sri City (No. IIITS/EC/2022/01) approved the proposed data collection protocol developed in accordance with the declaration of Helsinki and the “National Ethical Guidelines for Biomedical and Health Research involving human participants" of India.  Twenty-five healthy subjects with no history of upper limb pathology, including 22 males and 3 females, participated in the sEMG data collection process. The average age is 28 years. Before the first session of activities, each of the participants gave written informed consent and the data collection process is completely non-invasive.

Setup and acquisition protocol:

The 22 activities performed by each subject are listed in Table I. Each of the hand muscle activity is recorded with a 5-channel Noraxon Ultium wireless sEMG sensor setup. Five self-adhesive Ag/AgCL dual electrodes were placed at the center of the five most representative muscle sites of the right arm. The muscle locations are selected according to the atlas in chapter 17 of [2] and is given in Table II.  At the beginning of each session, the participant's hands are cleaned with an alcohol based wet wipe. Each subject is instructed to sit comfortably with one elbow resting on a table and an arm fixed at right angle compared to the forearm.

Prior to each session, the subject is acquainted with the experiment protocol including a video demonstration of the proposed activities. The total duration of each session is up-to one hour per subject depending on adaptability. Each activity is performed for a maximum duration of 10s and repeated 10 times. There is a rest period of 5s between each of the repetitions and a 30s gap between the sessions of different activities. Each of the activities consists of two phases: (1) an action and (2) rest. However, some of the activities included an extra release phase. During the action phase, the subject performs the corresponding activity; during the release phase, the subject transitions from the action state to rest  state; and during the rest phase, the subject completely relaxes each of his/her muscles. The general characteristics of the dataset are listed in Table III. Additional information is provided in tables IV to VI.  Tables are available in the EMAHA-DB1-Description document.

 Data files details:

The data is labelled for 25 subjects.

In the EMAHA-DB1, there are two folders - training data and test data. 

Since each activity has 10 trials, the trials [2, 5, 7] are separated into the test data. 

Within each of two folders, for each subject there is a mat file and its consists of the 9 columns of data 

 References:

1. N. K. Karnam, A. C. Turlapaty, S. R. Dubey, and B. Gokaraju, EMAHA-DB1: A New Upper Limb sEMG Dataset for Classification of Activities of Daily Living. arXiv, 2023. doi: 10.48550/ARXIV.2301.03325.   (please cite this paper)

2. Criswell, Eleanor. Cram's introduction to surface electromyography. Jones & Bartlett Publishers, 2010.

 The paper [1] also consists of a machine learning based analysis of FAABOS categories and the classification of 22 individual categories.