The effect of background muscle activity on computerized detection of sEMG onset and offset
The effect of background muscle activity on computerized detection of sEMG onset and offset
复制标题
DOI：
10.1016/j.jbiomech.2007.05.012
复制
发表时间：
2007-01-01
影响因子：
2.4
通讯作者：
Reeves, N. Peter
中科院分区：
文献类型：
--
作者：
Lee, Angela S.;Cholewicki, Jacek;Reeves, N. Peter
比较了两种计算机化的肌肉起始和偏移检测算法的性能。对常用的标准差(SD)方法和最近发展起来的近似广义似然比(AGLR)方法在不同的背景表面肌电信号(SEMG)活动水平下进行了评估。为此，在人工组装的肌电轨迹中，肌肉不活动时间和活动时间之间的幅度比从0.125到I变化。此外，在快速释放过程中，从各种躯干肌肉获得的1230个真实的sEMG信号被提升到3的幂，以改变它们的相对幅度比。随着背景活动的相对水平的增加，SD和AGLR方法都产生了更长的潜伏期和更少的肌肉反应，这表明如果被比较的受试者群体具有不同水平的背景肌肉活动，则可以引入检测伪像。在这两种方法中，AGLR似乎是受背景活动影响最小的。然而，在比率0.8以上，AGLR的结果也是不可靠的，特别是在检测偏移方面。接近此比率的平均延迟伪影：AGLR为8ms，SD为46ms。(C)2007爱思唯尔有限公司。保留所有权利。
The performance of two computerized algorithms for the detection of muscle onset and offset was compared. The standard deviation (SD) method, a commonly used algorithm, and the approximated generalized likelihood ratio (AGLR) method, a more recently developed algorithm, were evaluated at different levels of background surface EMG (sEMG) activity. For this purpose, the amplitude ratio between the period of muscle inactivity and activity was varied from 0.125 to I in artificially assembled sEMG traces. In addition, 1230 real sEMG signals, obtained from various trunk muscles during quick release, were raised to a power of 3 to change their relative amplitude ratio. As the relative level of background activity increased, both the SD and AGLR methods produced longer latencies and detected fewer muscle responses, suggesting that a detection artifact can be introduced if the subject populations being compared have different levels of background muscle activity. Of the two methods, AGLR appears to be the least affected by background activity. However, above the ratio 0.8, results from AGLR are also unreliable, particularly in detecting offsets. Average latency artifacts near this ratio were 8 ms for AGLR and 46 ms for SD. (c) 2007 Elsevier Ltd. All rights reserved.