Analysis of motor unit firing patterns in patients with central or peripheral lesions using singular-value decomposition

We applied the singular value decomposition (SVD) method to study single motor unit firing patterns. Two projects were carried out: (1) a computer simulation study to confirm the meanings of two SVD parameters, the eigenvalue corresponding to the positive-slope eigenvector (PEV) and that corresponding to the negative-slope eigenvector (NEV); and (2) a clinical study for which electromyographic (EMG) recordings were made from first dorsal interosseous muscle in patients with stroke, myopathies, or neuropathies and in healthy control subjects. Results of computer simulation reveal that the NEV reflects the amount of instantaneous firing variability, whereas the PEV/NEV (P/N) ratio exhibits the relative effect of a trend in the firing pattern. In human studies, the P/N ratio of stroke patients was significantly higher than that of the controls, whereas their NEV was comparable. By contrast, in the myopathy and neuropathy groups, the NEV increased significantly, whereas the P/N ratio did not. These results suggest that the SVD method decomposes the motor unit (MU) firing variation into two components and that the mechanism for increased firing variability is different for supraspinal and spinal-infraspinal lesions. (C) 2000 John Wiley and Sons, Inc.