SEP Waveform Classification via Median Frequency and Zero-Crossing Rate
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Somatosensory evoked potential (SEP) recordings are often contaminated with noise signals, which degrade the quality of the recordings and make it challenging to accurately measure latency and amplitude, thereby undermining the evaluation of neural structures in somatosensory pathways.
Innovation Solution
A method and system for classifying SEP recordings based on temporal and frequency characteristics, utilizing median frequency and zero-crossing rate as criteria to reject noise-affected recordings and enhance signal quality by averaging suitable recordings, thereby reducing the impact of noise signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If SEP recordings are taken to evaluate neural structures, then diagnostic information is obtained, but noise signals contaminate the recordings and degrade quality
Solution Approach 1:
The patent applies preliminary action by classifying SEP recordings based on temporal and frequency characteristics (median frequency and zero-crossing rate) before averaging them. This pre-classification step identifies and separates noise-affected recordings from clean recordings, ensuring that only high-quality recordings are averaged together, thereby preventing noise contamination in the final composite SEP recording.
2Reliability
If multiple SEP recordings are averaged to improve signal quality, then noise reduction is achieved, but the number of recordings required increases
Solution Approach 1:
The patent performs preliminary classification of each SEP recording using temporal and frequency characteristics before the averaging process. By identifying and selecting only the cleanest recordings (those with appropriate median frequency and zero-crossing rate characteristics) for inclusion in the average, the method achieves high signal quality with fewer recordings, rather than requiring a large number of recordings to statistically filter out noise.
Solution Approach 2:
The patent changes the parameters used for evaluating recording quality from traditional time-domain measures to frequency-domain parameters (median frequency) and temporal parameters (zero-crossing rate). These parameter changes enable more accurate identification of noise-free recordings, allowing selective averaging that achieves better signal quality with fewer recordings.
3Reliability
If traditional averaging of SEP recordings is performed, then some noise reduction occurs, but noise-affected recordings still degrade the composite recording
Solution Approach 1:
The patent applies preliminary classification based on median frequency and zero-crossing rate parameters before averaging. This pre-screening step identifies recordings with excessive noise or artifacts and excludes them from the averaging process. Consequently, the composite SEP recording is formed only from clean recordings, preserving measurement precision for latency and amplitude without the degrading effect of noise-contaminated traces.
Data Source
AI summary
Embodiments are disclosed relating to classification of somatosensory evoked potential waveforms.