IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. BME-28, NO. 7, JULY 1981 Characteristics of Somatosensory Evoked Potentials Recorded over the Spinal Cord and Brain of Man ARTHUR M. SHERWOOD, MEMBER, IEEE Abstract-Evoked potentials were recorded from the skin over the lumbar and cervical portions of the spinal cord, and the scalp over the sensory cortex of the brain, using averaging techniques. Responses could be identified over the cauda equina and root entry zone in the lumbar spine to stimulation of the tibial nerve at the popliteal fossa. These responses had characteristics of nerve root and spinal cord events in their thresholds, timing, duration, and refractoriness. Stimulation of the median nerve at the wrist likewise resulted in recognizable re- sponses over root entry portions of the cervical spinal cord. These later waves had a morphology suggestive of components arising from nerve plexus, nerve roots, and spinal cord. Responses recorded over the spinal cord were in the 1-10 ,uV amplitude range. Tibial, peroneal and median nerve stimulation were used to elicit 1-20 jV responses recorded over the cortex, which were found to be sensitive to the site, amplitude, and rate of stimulation. INTRODUCTION FOLLOWING early efforts at recording evoked potentials over the brain more than 30 years ago [1], the applica- tion of evoked potential recording has spread widely, to the point that it is currently used in virtually every medical center in this country. The availability of modern instrumentation, including high quality amplifiers and inexpensive digital signal processing equipment, makes it possible to meet the need for noninvasive monitoring of activity in the nervous system. Various applications of the technique for recording averaged evoked potentials have become useful in diagnosis, in longi- tudinal studies to detect regression or recovery of function, and to provide a means of monitoring the efficacy of the central nervous system during surgery on the spine or the spinal cord. Due to the small signals involved, recording of evoked neural activity has been more dependent on developments in elec- tronics than was, for example, recording of cardiac electrical activity, which contains sufficient energy for directly driving a recording device. Thus, progress from Berger's initial record- ing of spontaneous brain electrical activity [21, or electro- encephalography, to Dawson's work on averaging evoked po- tentials from the brain [1] was based, in part, on more than two decades of rapid progress in electronics. The earliest re- cordings from the human spinal cord by Magladery et al. [31 were made using invasive techniques, and depended on good amplifiers, not averaging. Modern computer technology has greatly facilitated the large amount of work being done today. Advances in the use of cortically recorded somatosensory Manuscript received January 13, 1981; revised March 4, 1981. This work was supported by the Bob and Vivian Smith Foundation, Hous- ton, TX, the Rehabilitation Research and Training Center No. 4 under Rehabilitation Services Administration Grant 16-P-56813-6, and the Rehabilitation Services Administration under Grant 13-P-59275-6. The author is with the Department of Clinical Neurophysiology, The Institute for Rehabilitation and Research, Houston, TX 77030. evoked potentials (SEP's) has been reviewed by Desmedt and Noel [4]. In the last two decades, noninvasive techniques for recording lumbar spinal cord evoked potentials (LSEP's) based upon signal enhancement through synchronous averag- ing have been developed by Liberson et al. [5 ], by Cracco [6], and others. More recently, Jones [7] and El-Negamy and Sedgwick [8] reported successful efforts to record and charac- terize cervical spinal cord evoked potentials (CSEP's) as well. The work reported in this paper has been done in the De- partment of Clinical Neurophysiology of The Institute for Re- habilitation and Research, a specialized rehabilitation hospital affiliated with the Baylor College of Medicine. Located within this hospital is a regional spinal cord injury center, as well as a center for treatment of patients with a variety of neuro- muscular and brain disorders. Within this setting, we are faced with the practical problems of restoring function to paralyzed individuals. In order to accomplish this goal, our own research is focused on the study of sensory and motor function, and particularly functions of the spinal cord in chronic patients with central nervous system lesions, varied in both site and extent. In this context, we use evoked potentials as one tool for exploring the site and degree of such lesions, along with a battery of other tests [9], [10]. We have focused upon re- cording from the lumbosacral portion of the spinal cord, from the cervical portion of the spinal cord, and from the sensory cortex of the brain, with stimulation of the peroneal and tibial nerves in the lower extremity, and the median nerve in the upper extremity. In this paper, we present tech- niques for recording these potentials and describe the charac- teristics of their waveforms. METHODS Data presented were collected from twelve male and five female normal healthy adults, ranging in age from 21 to 65 years, and from one T8 female spinal cord injury patient, 27 years old. Potentials were recorded using TECA AA6 Mk III amplifiers with a gain of 25 000, bandwidth of 16 Hz-8 kHz, and an in- put impedance of more than 100 MQ2. Amplifier noise re- ferred to the input was less than 2 ,V peak-to-peak. Samples were digitized with a Preston analog-to-digital converter which has a maximum digitization speed of 500 kHz and a +- 2 V full-scale range with a 12 bit plus sign digital output, for a resolution of 500 ,uV/bit. With a 16 channel multi- plexer, this allows a sampling rate of 31 kHz per channel, or up to 500 kHz for one channel. Data from the ADC are passed from a Hewlett-Packard 1000 data processing system with an HP 2111 F minicomputer and disk and magnetic tape storage. 0018-9294/81/0700-0481$00.75 C 1981 IEEE 481