Computerized Signature Discrimination of Above-Lesion EMG to Activate Functional Electrical Stimulation of Peripheral Lower Extremity Nerves of Complete Paraplegics

Abstract

The paper describes first patient results for a microcomputer based signature discrimination of raw EMC obtained by surface electrodes located at the erector spinae above the level of the lesion of a T-6 and of a T-10 spinal cord complete lesion (total) paraplegic. It is shown that 4 distinct and re-peatable signatures can be easily discriminated to activate all 4 basic functions (stand up, sit down, right-leg-up-then-down, left-leg-up-then-down) of primitive walking as produced by functional electrical stimulation (FES) of the intact peripheral nerves of the patient's lower extremities. The EMC signatures are each generated when the patient produces with his trunk above the lesion the phantom movement that would have produced the corresponding lower-extremity function of a non-paraplegic. Total activation delay is of 0.3 seconds, using a Motorola 68000 microprocessor. Signature discrimination is via comparing the parameters of a 4'th order pure-autoregressive statistical model of the raw EMC's time series, and its sample variance, this model being the one whose error relative to the true signal is white noise and hence free of linearly-retrievable information.The EMC signature discrimination system above has been interfaced with Dr. Kralj's FES system. Presently, the stand-up and sit-down functions have been successfully and repeatedly executed by one of the patients above via the EMC signature discrimination command of the FES system.