Classification of coma/brain-death EEG dataset based on one-dimensional convolutional neural network.

Area of Science:

• Neuroscience
• Medical Technology
• Signal Processing

Background:

• Electroencephalography (EEG) is crucial for diagnosing brain death in clinical settings.
• Intensive Care Unit (ICU) environments present challenges like electromagnetic noise and sedatives that can distort EEG signals.
• Accurate EEG interpretation is vital for timely and correct clinical judgments.

Purpose of the Study:

• To develop an efficient EEG signal pre-processing method and a classification system for coma and brain-death patients.
• To improve the accuracy and reliability of EEG-based diagnosis in challenging clinical environments.
• To assist physicians in making accurate judgments regarding brain death.

Main Methods:

• A band-pass filter and threshold rejection-based method was used for EEG signal pre-processing.
• A One Dimensional Convolutional Neural Network (1D-CNN) model was employed for classification.
• The system was trained and tested on real-world clinical EEG data from coma and brain-death patients.

Main Results:

• The proposed method achieved high classification accuracy (99.71%), F1-score (99.71%), and recall (99.51%) in distinguishing between coma and brain-death patients.
• The system demonstrated effective classification of informative brain activity features from noisy clinical EEG data.
• The experimental results validate the proposed model's performance in the coma/brain-death EEG signal classification task.

Conclusions:

• The study presents a straightforward and effective method for pre-processing and classifying EEG signals for brain death diagnosis.
• The developed system demonstrates validity and reliability in aiding physicians' diagnoses in complex clinical settings.
• This work has significant implications for constructing practical brain-death identification systems and choosing appropriate signal pre-processing techniques.