Application of fractal dimension for EEG based diagnosis of encephalopathy

In this study, we have investigated whether fractal dimension is a useful non linear feature for distinguishing electroencephalogram (EEG) of cases with encephalopathy from that of normal healthy EEGs. Both Higuchi's fractal dimension and Katz's fractal dimension were computed and were statistically analyzed between the normal and disease groups. Both parameters showed significant difference between the normal and encephalopathy groups, though Higuchi's fractal dimension showed better discriminating ability. Support Vector Machine (SVM) classifier was also applied for the automated diagnosis of encephalopathy based on EEG. It has been found that SVM classifier performed better when Higuchi's fractal dimension was utilized as feature set than using both Higuchi's and Katz's FD together.