MODELING OF EMG SIGNALS; MODEL-BASED INTERPRETATION OF ARRAY SIGNALS

Many anatomical and physiological parameters of a muscle are not accessible and cannot be measured directly. However, they are accessible in a model simulating EMG signals and EMG variables and may be changed until the simulated ob­servable EMG variables and parameters match the experi­mental ones. When the matching is obtained, it is likely that the parameters of the model have values similar to those that cannot be measured directly from the real system. This con­clusion must always be taken with caution since (a) a model always implies approximations and simplifications that may affect the results and (b) there may be more than one set of model parameters that provide a good fit of the experimental data. Motor unit action potential models have been developed by many researchers among which are P. Rosenfalck (41) and N. Dimitrova (42). The model described in Fig. 14(a) is based on the work of Gootzen et al. (43) and has been used to inves­tigate and explain some experimental findings. An application example is provided by Fig. 14(b), which shows 10 firings of the same MU of a healthy biceps brachii during a low-level voluntary contraction. The signals are detected bipolarly from a 16-contact linear array.

(a)

(b)

Figure 13. Initial 300 ms of EMG obtained from bipolar differential measurement with elec­trodes over biceps and triceps during elbow flexion. (a) Four 300 ms records and (b) the ensemble average of sixty 300 ms records demonstrating the deterministic component of the initial phase [from Hudgins et al. (35)].

The 10 firings are selected during a time interval of 1.5 s, are aligned and superimposed, and are similar enough to jus­tify the assumption that they belong to the same MU. The results of the simulation (open circles) are superimposed and the indicated model parameters provide an estimate for ana­tomical features of the MU, conduction velocity, and anisot­ropy of the tissue. Future research might lead to the develop­ment of systems for the automatic identification of the most likely set of parameters for individual MUs and make them available to the neurologist for diagnostic evaluation.

Conductive medium: ay? az; ay = ax

(a)

Electrode pair

: exp. data

о : simulation Model parameters:

e = 10 mm a, /a. = 6 b = 7 mm a/b = 1/3 h = 4.5 mm R = 2 mm W| = 5 mm WTR = 20 mm WTL = 10 mm LR = 56 mm Ll = 73 mm

0 5 10 15 20 25 CV = m/s

Time (ms)

(b)

Figure 14. (a) Model for the simulation of surface EMG signals and of their variables. Schematic structure of the model of a single motor unit. The motor unit has N fibers uniformly distributed in a cylinder of radius R at depth h. The axis of this cylinder may present an angle with respect to the skin plane and with respect to the z axis. The neu­romuscular junctions are uniformly distributed in a region WI, and the fiber-tendon terminations are uniformly distributed in two regions WTR and WTL. A right and a left current tripole originate from each neuromuscular junction and propagate to the fiber-tendon termination, where they be­come extinguished. The conduction velocity is the same in both directions and for all fibers of a mo­tor unit but may be different in different motor units. Each of the voltages VA, VB, VC, and VD is the summation of the contributions of each tri­pole. (b) Example of simulation of 10 superim­posed firings of a motor unit detected during a low-level contraction of a healthy biceps brachii muscle with a linear 16 contacts array. Pair 14 is proximal, pair 0 is distal.

Добавить комментарий

Ваш e-mail не будет опубликован. Обязательные поля помечены *