CORRELATION OF DEVELOPMENTAL NEUROLOGICAL FINDINGS WITH SPECTRAL ANALYTICAL EEG EVALUATIONS IN PRESCHOOL AGE CHILDREN

For the differentiation of developmental neurological disorders in pre -school age children, the relationship between automatically derived E EG parameters and developmental neurological findings was investigated . Within the scope of the Munich Pediatric Longitudinal Study, the sam ple sets of 4- and 5-year-old children (according to the frontal and p arieto-occipital EEG derivations) with selected abnormal findings cate gorized by special items were compared with the corresponding control groups. This was carried out by means of one-sided t tests and relativ e frequency band-related as well as single-step spectral power paramet ers in the alpha range of the EEG. Automatic analysis using single-ste p power values was superior to that using band-related parameters. Thi s led to the conclusion that use of age-specific single step parameter s for a quantitative EEG analysis and ignoring the classical frequency bands will yield statistically greatly improved results. For 4- and 5 -year-old children, the best separation of the neurologically abnormal groups from the normal control groups was obtained using relative spe ctral values in the frequency range of 9.0-9.8 Hz with a maximum at 9. 4 Hz. At the same time, the topographical conditions of brain immatura tion should be taken into account. The results for the children examin ed in this study differ in a stronger distinction over the frontocentr al brain region of 4- and 5-year-olds (P < 0.01) and through an additi onal distinction over the parietooccipital region of the 5-year-olds ( P < 0.001). It still must be tested whether the spectral parameter at 9.4 Hz is age-specific for 4- and 5-year-old children or whether in ot her age groups different spectral parameters are of use. As an examine r-independent method, the automatic EEG analysis should become an inte gral component of developmental neurological diagnostics. (C) 1997 Els evier Science Ireland Ltd.