Study on Adaptive Noise Canceller Based on Fixed-Point Algorithm for Real-Time Somatosensory Evoked Potential Monitoring

The objective of this study was to analyze the performance of fixed-point adaptive noise canceller (ANC) for SEP measurement, as well as compare with the outputs of floating-point ANC at 16 bits word-length. ANC technique has been reported powerful to improve signal to noise ratio (SNR) of somsatosensory evoked potential (SEP). For the sake of real-time monitoring of intraoperative SEP, both fixed-point and floating point algorithms can be applied to Field Programmable Gate Array (FPGA) design. However, the ability of fixed-point ANC to improve the SNR of SEP is still dubious. In simulation, SEP signals from 7 patients were included. The results showed that outputs of fixed-point ANC had higher distortion index (DI) from real SEP signals than that of floating-point ANC at the same step-size parameter (mu), and optimal range of mu value was different in fixed-point algorithm from floating-point algorithm at the same DI. However, the two algorithms had very similar DI trend curves when mu value and trail number varied in a certain range, the fixed-point ANC can extract SEP from raw signals as effectively as floating-point ANC, compared with ensemble averaging SEP of 300 raw SEP signals. It would be helpful to design fixed-point ANC based on FPGA for real-time monitoring of intraoperative SEP.