Decrease of non-linear structure in the EEG of Alzheimer patients compared to healthy controls.

OBJECTIVE: Non-linear EEG analysis can provide information about the functioning of neural networks that cannot be obtained with linear analysis. The correlation dimension (D2) is considered to be a reflection of the complexity of the cortical dynamics underlying the EEG signal. The presence of non-linear dynamics can be determined by comparing the D2 calculated from original EEG data with the D2 from phase-randomized surrogate data. METHODS: In a prospective study, we used this method in order to investigate non-linear structure in the EEG of Alzheimer patients and controls. Twenty-four patients (mean age 75.6 years) with 'probable Alzheimer's disease' (NINCDS-ADRDA criteria) and 22 controls (mean age 70.3 years) were examined. D2 was calculated from original and surrogate data at 16 electrodes and in three conditions: with eyes open, eyes closed and during mental arithmetic. RESULTS: D2 was significantly lower in the Alzheimer patients compared to controls (P = 0.023). The difference between original and surrogate data was significant in both groups, implicating that non-linear dynamics play a role in the D2 value. Moreover, this difference between original and surrogate data was smaller in the patient group. D2 increased with activation, but not significantly more in controls than in patients. CONCLUSIONS: In conclusion, we found decreased dimensional complexity in the EEG of Alzheimer patients. This decrease seems to be attributable at least partially to different non-linear EEG dynamics. Because of this, non-linear EEG analysis could be a useful tool to increase our insight into brain dysfunction in Alzheimer's disease.

Diagnostic usefulness of linear and nonlinear quantitative EEG analysis in Alzheimer's disease.

The sensitivity of the EEG in early AD is somewhat limited. In this respect spectral analysis is little better than visual assessment. In this study we address the question whether a new type of EEG analysis derived from chaos theory can improve the sensitivity of the EEG. EEGs were recorded in 15 control subjects and 15 patients with mild AD. The EEG recorded at 02 and 01 during eyes closed and eyes open conditions was subjected to spectral analysis (relative power) and nonlinear analysis (calculation of the correlation dimension D2). AD patients had more relative theta power and impaired reactivity in alpha, delta and theta bands. Also, reactivity of the D2 was impaired in AD subjects. For a specificity of 100%, relative theta power had the highest sensitivity (46.7%). Alpha band reactivity at O1 had a sensitivity of 40% and D2 reactivity at O1 had a sensitivity of 33.3%. Combining theta power with alpha reactivity resulted in a sensitivity of 53.3%; combining theta with D2 reactivity resulted in a sensitivity of 60%. Used in isolation, linear analysis was superior in differentiating AD patients from controls. The best results were obtained by combining linear with nonlinear measures. This approach does not seem practical yet, but deserves further study.