EMG-based analysis of change in muscle activity during simulated driving

Summary

Extensive usage of computers could cause fatigue and even lead to musculo-skeletal injuries. Onset of fatigue can be determined by analyzing biosignals like surface electromyogram (sEMG), electrocardiogram (ECG), electroencephalogram (EEG) and electrooculogram (EOG). Physiological parameters such as pressure, respiration rate and oxygen debt have also been used to estimate the onset of fatigue and recovery time. Other than these measures, behavioral analysis using video and subjective analysis have also been used extensively to detect onset of fatigue. In this study, muscle activity changes in the shoulder and neck muscles while gaming in an automobile simulator, using sEMG, are analyzed. Two groups of professional and non-professional drivers participated in this study. Statistically significant (p<0.05) change in muscle activity is found in both the groups during a short duration (15 min) of gaming. Such analysis can be used to identify weak muscles in subjects, which can be used as an input for necessary rehabilitation or training programs.

Introduction

A large number of young people utilize personal computer as a gaming console. Most games are usually target oriented and require the user to perform repetitive motions in a confined space. Added to this, pressure of time and target, forces the user to focus more on the game than their physical well being. Computer-related work often involves repetitive motions of certain muscles and joints. Long periods of such an exposure can lead to musculo-skeletal disorders (MSD) (Srinivasan and Balasubramanian, 2005; Varadhan et al., 2005). Low back, neck and shoulder pain are common syndromes among long-time computer users (De Wall et al., 1992). Relationship between fatigue and pain in muscles has been extensively reported in literature (Westgaard, 1999; Nilsen et al., 2006). Hence, detection of muscle fatigue can be useful in understanding muscular pain and related disorders. Electroencephalogram (EEG), electrocardiogram (ECG), surface electromyogram (sEMG), oxygen debt and breathing patterns are commonly used to detect such changes (Schier, 2000).

EEG is a non-invasive technique used to measure electrical activity of the brain (Lin et al., 2005). EEG frequency bands namely alpha, beta, theta and delta are manifestations of a particular mental state. Hence, corresponding changes in an EEG frequency band can be a useful indicator of onset of fatigue (Trejo et al., 2004). It has been reported that alpha rhythms attenuate, diminish and reappear after few seconds during drowsiness, which finally disappear during sleep (Kecklund and Akerstedt, 1993). Hence, a change in alpha rhythm could be used as an indicator of sleep (Lafrance and Dumont, 2000). Another study reports that drivers experience micro-sleep, which is accompanied by changes in alertness (Harmony et al., 1993). Monitoring alertness during driving can be useful in detecting fatigue and hence could be used as feedback to avoid accidents. In another study involving simulated driving, an increase in occurrence of theta and delta waves was found during micro-sleep (Lal and Craig, 2002).

However, limitation of these studies is the reproducibility of similar changes in EEG, which was verified by correlating randomly selected EEG parts related to micro-sleep periods (Lal and Craig, 2005). Selected periods were highly correlated, which suggest that the study could be used for setting up a real-time driver fatigue detector. EEG can be used to determine only cognitive fatigue, which is a manifestation of cranial activity and can be effective only if it is correlated with measures of physical fatigue.

Physical fatigue can be indirectly detected by measuring oxygen debt. Onset of fatigue was subjectively analyzed among 10 drivers using levels of oxygen supplied to them while driving in a simulated environment (Sung et al., 2005). They reported that subjective feeling of fatigue was high at 18% oxygen concentration and low at 30% oxygen concentration level. Another method of determining oxygen debt is by measuring blood lactate level after regular intervals of time (Ozturk et al., 1998). The results reported that muscle could retain normal strength only after 10 min of rest. But blood lactate measurement is invasive in nature and hence alternative ways of identifying fatigue should be adopted.

Fatigue can occur because of excessive workloads and stress, which can be detected by variations in heart rate. Temporal and frequency indices obtained by analyzing ECG were used to detect fatigue in seven middle distance runners and a sedentary group. The study reported a significant decrease in parasympathetic indices up to 3 weeks of heavy training, which was followed by a significant increase in the same index during recovery period. In another study workload, stress and fatigue encountered by anesthesiologist during their work were monitored using Holter ECG and a questionnaire (Kain et al., 2002). They reported that the graphs and workload scores obtained as a result of analyzing ECG can be useful in analyzing fatigue, stress and workload of anesthesiologists (Pichot et al., 2000).

Blood lactate level is an indirect invasive technique of measuring physiological muscle fatigue, whereas, reproducibility is a major issue in analysis of the EEG and electrooculogram (EOG). Moreover, to comprehend fatigue using EEG and EOG, a synchronized analysis of behavior such as eye movement, change in facial expression, etc. is required. It is also well known that EEG and EOG are measures of cognitive fatigue, which can lead to changes in muscular activity. Similarly heart rate variability is also an indirect method of measuring fatigue.

The sEMG is a non-invasive index of the level of muscle activation (De Luca, 1984) and hence can be used directly to identify weak muscles. Strengthening the weak muscles by a rehabilitation program can help the subjects to perform better. Fast Fourier transforms (FFT) is the most popular signal processing technique used for analyzing sEMG. Fatigue detection using sEMG signal is associated with change in mean frequency (MF) (Luttmann et al., 1996; Balasubramanian and Prasad, 2006), median frequency (Luttmann et al., 2000) and time domain parameters like RMS amplitude (Petrofsky, 1979), integrated electromyogram (iEMG) (Weir et al., 1998) and electrical activation (Lina et al., 2004). A wavelet technique is a good technique for fatigue detection, reaction time detection or pattern recognition for sEMG signal (Karlsson and Gerdle, 2001). Hostens et al. (2004) and Kumar et al. (2003) reported that wavelet transform is a better alternative technique for analyzing sEMG than Fourier transform since it can be used without loss of information in non-stationary conditions. The objective of this study is to determine the changes in shoulder and neck muscle activity of professional and non-professional drivers during simulated driving by analyzing shorter duration sEMG recording, which could be helpful in identifying the weak muscles.