MYOELECTRIC SIGNAL ANALYSIS IN HANSENOLOGY

Abstract

The human body is replete with electrical activity. The development of knowledge on the electrophysiology compounded by an exponential growth in instrument technology has made it possible to trap the electrical potentials generated by the human body. That in turn can be amplified, stored, analyzed and the results applied to benefit mankind. Of the multitudinous variety of bioelectric signals known to man, this thesis has its focus on the myoelectric signals. The field of electromyography itself is wide and evergrowing and the better known among them are the foursome; prosthetics, orthotics, robotics and diagnostics. This treatise is confined to the diagnostic aspects of electromyography. The myoelectrical signals can be let off via a surface electrode or a needle electrode. Signals picked up by surface electrodes are eminently suitable for the first three areas. For diagnostics, needle electrodes have been almost exclusively used. Of late, lot of research efforts have gone into making surface electromyography a viable tool for diagnostics. Once its usefulness has been conclusively shown, its acceptance will be universal, because of the clear cut advantages (the reasons are (i) it is noninvasive and consequently painless, (ii) it enables large areas to be investigated in a single effort (about twenty times as much as with a needle electrode), (iii) it yields reliable parameters that can be used to quantify the neuromuscular disorders on the level of individual muscles, and

(iv) routine tests with surface EMG can be performed by nonclinicians). Hansenology is a study of Hansen's disease(HD). Widely known as leprosy, its earliest irrefutable records date as far back as 600 BC. However, it was only in 1893, its cause - the neurotropic bacillus M. leprae - was identified. Essentially a disease of the peripheral nerves and skin. A study of this disease is of great pertinence to this country as every fifth among the estimated twenty millions afflicted worldwide is an Indian. The Government of India has rightly set 2000 AD as the target year for wiping out this scourge from this country. Since the introduction of the sulfones in 1941, to combat HD, research in chemotherapy has made such progress that the disease can now be arrested at any stage. Reconstructive surgery followed by physiotherapy can restore some degree of function to paralyzed extremities, remove some of the stigmata of the old disease and help regain cosmetic appeal. However, neither chemotherapy nor surgery can replace a lost toe or finger nor can they regenerate a damaged nerve. The former causes an ex patient to continue suffer social stigma and disability and the latter makes him susceptible to injury and trauma despite the elimination of the bacilli from his body. Early detection and adequate treatment can forestall both. There lies then a strong case for developing techniques for early detection. As peripheral nerves are always involved in HD and there is no non-neural HD, the nerve involvement causes maximum damage to a (v) patient and the bacillus' predeliction for nerves makes HD unique, the electrical evaluation of nerves through the associated muscles may hold the key to early diagnosis. Also it is opined that electromyographic tests frequently turn positive prior to the appearance of clinical symptomatology of neural deficits. A review of relevent literature shows that the electrodiagnostic tests in HD are mostly confined to the evaluation of the nerve conduction velocity (NCV) both motor and sensory. The NCV test suffers from the fact that there is a wide variation in its value even among the healthy. This is a pointer to the fact that some other areas need to be probed, to hone the diagnostic prowess of electromyography, taking recourse to the modern analytical tools. This thesis is an effort to explore new grounds for a better understanding of the electropathophysiology of HD leading to a precocious diagnosis, taking myoelectric signal as the basis. Myoelectric signals are of stochastical manifestation, in the sense that the repeat records of the same signal under identical conditions would have only statistical characteristics in common with the original. Statistics being the only tool for analyzing and interpreting such signals, this thesis aims at investigating them as statistical entities. The work can be divided into three sections. (i) Time domain analysis of ME signals (ii) Frequency domain analysis of ME signals (iii) ME signal simulation and microprocessor

based FFT analysis. For the first two studies the real ME signals of both normal and HD subjects recorded at a leprosy research centre are used. The paper recorded signals are sampled and the data are stored in a main frame computer or an IBM compatible PC in separate data files for each record. To start with, an attempt was made to analyze the signal in that domain itself as biosignals are invariably available in that domain. Parameters of significance such as the maximum potential, maximum slope and the number of zero crossings, all in a given duration are extracted using simple algorithms. Using this data, the correlation between any two of the aforesaid parameters is computed separately for the normal subjects and the HD patients. The patterns are observed to be distinctly different for the two categories. The reasons for the same are elaborated and interpreted. In the next section, the signals are analyzed in the frequency domain using the well known Fast Fourier Transform (FFT) algorithms. The raw data here is filtered using convolution techniques incorporating a Hamming data window. The filtered data is then frequency transformed using modified versions of the algorithms and the log power spectrum so obtained is plotted using a CALCOMP plotter as well as a dot matrix printer. The plots show the frequency range where the normal subjects score over the HD afflicted. Knowing this spectrum, the average power over this region is computed separately for the healthy and the (vii) diseased. The results are tabulated. Also plotted alongside is the estimate of the autocovariance. The plot depicts the random nature of the the healthy signal and its gradual loss in the diseased as the nerve condition deteriorates. This corroborates the observations made in the time domain analysis. The third and the last section has two segments. The first segment deals with the development of simulated ME signal that is TTL compatible. By virtue of its random nature the ME signals contains components of all frequencies ranging from 2Hz to 20KHz. The experiment aims at synthesizing the interference pattern representing a surface signal using simulated action potentials. In the second segment, an ADC circuit and the associated software for digitizing a signal and storing it in the expanded RAM area of a microprocessing system is developed and tested. Subsequently some of the useful subroutines for FFT analysis of a signal are] developed and tested. The object is to evaluate the suitability of microprocessors for on-line analysis of real ME signals. The. fast development of microprocessor technology has made available increasingly powerful systems at ever decreasing costs. Hence this effort may pave the way for envisaging a portable microprocessor based instrument that a field worker among the HD afflicted may carry along for determination of factors of diagnostic and prognostic value.