Wiener filtering and classification of neurographic recordings

Methods were developed to improve the S/N ratio and selectivity of nerve recordings, and to allow classification of nerve signals originating from activation of different fiber diameter populations (e.g. different receptors). The need for these methods arose from our ´bladder pacemaker´ research, where nerve cuff electrodes were placed around the feline sacral spinal roots to monitor urinary bladder afferent activity. A novel technique to design Wiener filters for neurographic recordings was developed, which resulted in improved S/N ratio and selectivity of the recordings. Since the sacral roots innervate dermatomes and the rectum, the nerve signal increases had to be classified to extract bladder afferent information only. Multilayer perceptron neural network was trained with different autocorrelation functions to classify bladder, rectal and cutaneous nerve signal increases.