径向基函数自适应减法器提取诱发电位的性能评价

Performance evaluation of adaptive noise canceller based on radial basis function for evoked potentials extraction

目的:针对体感诱发电位(somatosensory evoked potentials,SEP)的特征,基于现场可编程门阵列(field pro grammable gate array,FPGA)硬件平台设计径向基函数自适应减法器,实现体感诱发电位的快速提取。方法:通过对各模块的硬件算法设计,利用Simulink仿真工具对5例接受脊柱侧弯手术患者的原始SEP信号进行仿真实验,以工频干扰、EEG脑电为噪声,对径向基函数自适应减法器提取SEP的性能进行评价。结果:不同输入信噪比条件下,径向基函数自适应减法器比单一自适应减法器输出信号的信噪比高;输入信噪比为-15 dB时,单一的自适应噪声减法器(adaptive noise canceller,ANC)输出相对于模板信号的失真比径向基函数自适应减法器(ANC-RBF)小,但输入信噪比为-25、-30 dB时,ANC输出相对于模板信号的失真比ANC-RBF大;ANC-RBF提取的SEP波形比较平滑。结论:径向基函数自适应减法器比单一自适应减法器对术中体感诱发电位具有更好的去噪效果,不仅提取信号的失真度较小,而且信号波形更为平滑,使SEP信号的潜伏期和幅值更易识别。

Objective To design a radial basis function adaptive noise canceller （ANC-RBF） based on FPGA according to the characteristics of somatosensory evoked potentials （SEP） for the purpose of conducting fast detection of SEP.Methods To design the modules of FPGA, power frequency interference and EEG were considered as noise signal, original data from 5 patients undergoing scoliosis surgery was processed by Simulink, and the SEP detection effect of ANC-RBF was evaluated. Results Signal to noise ratio （SNR） of the outputs from ANC-RBF were higher than that of mono ANC processor under different SNR inputs. When SNR was -15 dB, the distortion of output from ANC to module was smaller than that of ANC-RBF. The distortions of output from ANC to modules were higher than that of ANC-RBF when SNR were -25 and -30 dB, respectively, and the SEP signals detected by ANC-RBF were relatively smooth.Conclusion ANC-RBF has better noise cancellation effect on detection of SEP over that of mono ANC. The distortion of the detected SEP is small, and SEP signal is smooth, which make it easy to distinguish the amplitude and latency of the SEP signals.