Methods to improve the signal-to-noise ratio during the recording of biological signals

A somatosensory evoked potential (SEP) is electrical activity measured at the skin surface in response to nerve stimulation. SEPs can indicate problems with signal pathways in the central nervous system such as insults during spinal cord surgeries. However, current SEP monitoring provides delayed feedback, which increases the risk of critical neurological damage during surgery.

The team designed a software algorithm that enhances current SEP monitoring with efficient signal acquisition and analysis. This improves the signal-to-noise ratio (SNR) and provides faster feedback, enabling early intervention to prevent permanent nerve damage.

The algorithm has four blocks: 1) the data collection block processes raw signals and discards corrupted trials based on amplitude thresholds, 2) the model creation block develops an autoregressive model to accurately predict SEP signals, 3) the filtering block uses this model to filter out noise from relevant SEP signals in the recorded data, and 4) the display block calculates the SNR and displays the filtered data.

The system goal is to provide the surgeon with complete feedback within 20 seconds. The current state-of the-art for SEP monitoring is 3 minutes. The SNR will be improved by >= 20db, and tests will ensure the system has 90% sensitivity and 95% specificity. Ultimately, this will significantly reduce postoperative complications with more accurate, real-time feedback during surgeries.