Premium
Outliers, a way to detect abnormality in quantitative EMG
Author(s) -
Stålberg Erik,
Bischoff Christian,
Falck Björn
Publication year - 1994
Publication title -
muscle and nerve
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.025
H-Index - 145
eISSN - 1097-4598
pISSN - 0148-639X
DOI - 10.1002/mus.880170406
Subject(s) - abnormality , outlier , motor unit , pattern recognition (psychology) , medicine , mathematics , statistics , artificial intelligence , computer science , anatomy , psychiatry
Abstract In visual analysis of motor unit potentials it is common to decide abnormality by a few motor unit potentials with definitely abnormal amplitude, duration, and shape. The aim of the present investigation was to define limits of normal values and to compare the diagnostic yield of assessing definitely abnormal values outliers, with conventional mean values of MUP parameters. MUPs were extracted and measured with a new decomposition method. Reference values were obtained for three commonly studied muscles. Patients with various types of neuropathies and myopathies were studied in the same way with measurement of outliers and mean values. It was found that outliers were as sensitive as mean values in neuropathies and better in myopathies. Often an increased number of outliers could already be detected after only a few MUPs had been obtained. It would not have been necessary to obtain all 20 MUPs in these patients. The conclusion is that the outlier method is as sensitive as mean values. Because the number of MUPs required may be reduced, the investigation takes a shorter time and is less painful for the patient. If the degree of abnormality is to be quantified, calculation of mean values is still necessary. The combination of outliers and mean values may be the optimal way to detect and express abnormality. © 1994 John Wiley & Sons, Inc.