An Integrated Spatial-Spectral Denoising Framework for Robust Electrically Evoked Compound Action Potential...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Neuroscience

Background:

• Electrically evoked compound action potential (ECAP) is vital for assessing auditory nerve function.
• Accurate ECAP analysis, especially with the panoramic ECAP (PECAP) framework, relies on clean signals.
• Noise, particularly in low signal-to-noise ratio (SNR) conditions, compromises ECAP parameter estimation.

Purpose of the Study:

• To develop and evaluate a novel two-stage preprocessing denoising (TSPD) algorithm for enhancing ECAP signals.
• To improve the accuracy of auditory neural activity patterns and ECAP magnitude estimation in noisy environments.
• To introduce an integrated spatial-spectral denoising framework, PECAP-TSPD, for robust ECAP analysis.

Main Methods:

• Constructed an ECAP matrix using forward-masking, treating the signal as a 2D image.
• Applied an improved spatial median (I-Median) filter for spatial noise reduction.
• Utilized a log-spectral amplitude (LSA) Wiener filter for spectral domain denoising on the vectorized matrix.

Main Results:

• The proposed PECAP-TSPD method demonstrated superior performance in objective quality measures.
• Achieved the lowest average RMSE for PECAP magnitudes (1.952%) and neural patterns (1.407%).
• Obtained the highest average TDCC (0.9988) and SSIM (0.9931) compared to existing methods under impulse noise.

Conclusions:

• The PECAP-TSPD algorithm effectively denoises ECAP signals, outperforming PECAP, PECAP-CNN, and PECAP-I-Median.
• This spatial-spectral denoising approach significantly enhances the reliability of ECAP-based auditory nerve evaluation.
• The findings suggest PECAP-TSPD is a promising tool for clinical applications requiring precise ECAP analysis.