1514 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 54, NO. 8, AUGUST 2007 Differences in the Activity of the Muscles in the Forearm of Individuals with a Congenital Absence of the Hand Sean D. Taffler and Peter J. Kyberd* Abstract—The spectral content of the myoelectric signals from the muscles of the remnant forearms of three persons with con- genital absences (CA) of their forearms was compared with signals from their intact contra-lateral limbs, similar muscles in three per- sons with acquired losses (AL) and seven persons without absences [no loss (NL)]. The observed bandwidth for the CA subjects was broader with peak energy between 200 and 300 Hz. While the sig- nals from the contra-lateral limbs and the AL and NL subjects was in the 100–150 Hz range. The mean skew of the signals from the AL subjects was and those with NL of , while the signals from those with CAs had a skew of . The structure of the muscles of one CA subject was observed ul- trasonically. The muscle showed greater disruption than normally developed muscles. It is speculated that the myographic signal re- flects the structure of the muscle which has developed in a more disorganized manner as a result of the muscle not being stretched by other muscles across the missing distal joint, even in the muscles that are used regularly to control arm prostheses. Index Terms—Congenital absence, muscle, prosthetics, spectral analysis, surface EMG, ultrasound. I. INTRODUCTION F OR surface electrodes each electromyographic (EMG) signal depends on the summation of a large number of individual voltage spikes from the fibers that constitute muscle [1], [2]. The resulting signal depends on the electrode size, shape, and position [3]. For any specific experimental setup, the signal reflects the effect that the intervening flesh will impose on the signal, be it muscle, skin, or subcutaneous fat [4], [5]. Thus, for any given experimental configuration the signal reflects the structure of the muscle. Electromyograms are commonly used for diagnostic pur- poses, detecting changes in the muscle due to disease, or to sense muscle activity when the subject is being monitored, they may also be used as the command input to hand prostheses. In this form, the signal is generally rectified and smoothed so that the signal approximately follows the level of contraction of the muscle. Other paradigms are also used [6], but simple rectifi- cation and level-based control are most common. This allows the level of the contraction to be sufficiently well controlled by Manuscript received April 26, 2006; revised December 6, 2006. This work was supported in part by The Wishbone Trust. Asterisk indicates corresponding author S. D. Taffler was at Hertford College, Oxford University,Oxford, U.K. *P.J. Kyberd was in the Department of Cybernetics, Reading University, Reading, U.K. He is now in the Institute of Biomedical Engineering, 25 Dineen Drive, PO Box 4400, University of New Brunswick, Fredericton, NB E3B 5A3, Canada. (e-mail: pkyberd@unb.ca) Digital Object Identifier 10.1109/TBME.2007.900817 the user and easily interpreted by the controller, to allow joint or limb control that is acceptably close to proportional, [7], [6]. Limb absence results from two basic mechanisms: The first is due to anomalies in foetal development, the second is the re- sult of some form of amputation. For prosthetic supply, the two groups are generally treated as entirely interchangeable. The re- sult is that, while many other disorders have been classified from the content of their EMG signals, only one EMG processing system is employed for all prostheses users, and the general at- titude is that they can be treated as an homogenous group. This preliminary study shows a limitation of this assumption. Observation of the raw EMG signals from some prosthesis users showed that the signals appeared to be different, closer study suggested that some of the signals possessed a broader bandwidth than reported in the literature. Thus, this study was initiated. To facilitate the measurements, an amplifier with a wider bandwidth than their conventional product was commis- sioned from Delsys. II. METHOD A. Subjects Three individuals who had a congenital absence (CA) of one of their forearms were recruited from the population of users attending the Oxford Limb fitting clinic at the Nuffield Orthopaedic Centre, U.K. The subjects were those who fitted the selection criteria, could generate a voluntary signal that could be recorded and were willing to volunteer their time for conducting the tests. Potential subjects were selected carefully to ensure sufficient uniformity in the range and type of loss. Thus, each subject was a long term habitual user of a prosthesis, and were regarded as a regular users of their devices, as determined by the clinic team [8], (based on frequency of attendance of the clinic and the fre- quency that their prostheses needed maintenance). User surveys suggest that if a person uses their prosthesis they will employ it for the majority of the day (greater than 8 h) [9]. Hence, they used their arm and muscles under study regularly, either to drive the prosthesis or to position it in space for effective use. The muscles chosen were as distal as possible, below the ole- cranon. Thus, for those with a CA the muscles studied had never been attached to a working body segment. A second group con- sisted of three individuals who had lost a hand and wrist through amputation as adults [Acquired Loss (AL)]. All were recruited as part of a larger study on EMG control of prostheses 1 . Again 1 Ethical approval number C98–158—Oxford Regional Health Authority Ethics Committee) 0018-9294/$25.00 © 2007 IEEE