Estimation of the nerve conduction velocity distribution using partial bicepstral information

Abstract

Recording of the compound action potential (CAP) of a peripheral nerve is an essential procedure in clinical neurophysiology. The CAP is a linear summation of single fiber action potentials (SFAP's) propagating along the nerve fibers, and in the discrete time, it can be expressed as the circular convolution of a delay sequence (DS) and the sampled SFAP. The distribution of the conduction velocities (DCV) (related to the DS), and both amplitude and shape of the SFAP are estimated and can be used as a measure of health and well-functioning of the nerve. In the presented method, the blind deconvolution scheme proposed by Hirose is evaluated in the bicepstral domain to estimate the DS at a given conduction distance l 1 . The Bispectrum Iterative Reconstruction Algorithm (BIRA) is used to recover the delay sequence by only partial (phase) information. Linear phase is recovered by using the delay phase cepstrum, and the DCV is calculated from the estimated DS, following the formulation of the forward problem. The efficiency of the method is tested, especially, in the case of low SNR recordings.