EEG-based cerebral pattern analysis for neurological disorder detection via hybrid machine and deep learning approaches

Comprehensive Summary

This study combines a feature-based random forest (RF) machine learning model with an image-based convolutional neural network (CNN), creating a hybrid RF-CNN model to detect neurological disorders based on electroencephalography (EEG) signals. Disorders tested for included mild cognitive impairment, Alzheimer’s disease, and epilepsy. Inputs for analyzing cerebral patterns were generated from power-based features, spectral topographic maps, and continuous wavelet transform based scalograms. Based on the F1-scores, the RF-CNN model performed the best at accurately decoding what neurological disorders corresponded to specific EEG patterns, with an accuracy of 99.19% and an F1-score of 98.32%. The authors posit that the RF-CNN model is more accurate because it integrates both features and images.

Outcomes and Implications

The hybrid RF-CNN model is not only novel but also presents a more reliable analysis of transitional phases and non-linear brain dynamics, and so could be used to detect neurological disorders.