Complexity Analysis of EEG Signal in Patients with Cognitive Impairment Using the Hjorth Descriptor

Elderly people with mild cognitive impairment earlier, have a higher potential to develop Alzheimer's. This degenerative process that tends to be progressive can be avoided by proper treatment at the initial onset. One of the low cost medical modalities for this symptom analysis is electroencephalogram (EEG). Spectral analysis is considered capable of describing pathological conditions and is normally characterized by the slowing of the EEG signal. However, bias can occur because noise artifacts have a dominant power at low frequencies. Therefore, this research proposes an approach based on complexity analysis for the differentiation of subjects with cognitive impairment and normal subjects as an attempt to early detection of Alzheimer's. This analysis can be used as a support for traditional spectral analysis so that it increases accuracy. The study was conducted on 27 subjects consisting of 16 normal and 11 MCI patients who were recorded using 19 channel EEG. The Hjorth descriptor was applied to calculate complexity parameters in both EEG wave groups. Statistical analysis was applied to find significant differences. Results showed that MCI groups have lower complexity values than normal in almost all channels. Significant differences was found in F3 and Fz channels (p <; 0.05) compared with averages of all channels.