Quantitative Electroencephalography Analysis for Improved Assessment of Consciousness Levels in Deep Coma Patients...

Area of Science:

• Neuroscience
• Medical Technology
• Signal Processing

Background:

• Assessing a patient's level of consciousness (LeOC) is crucial in neurological evaluations.
• The Glasgow Coma Scale (GCS) is a standard, yet subjective, tool for neurological assessment.
• Objective methods are needed to evaluate consciousness levels, especially in deep coma states (GCS 3-8).

Purpose of the Study:

• To objectively evaluate the Glasgow Coma Scale (GCS) using electroencephalography (EEG) signals.
• To analyze EEG features for differentiating levels of consciousness in deep coma patients.
• To develop a machine learning model for classifying GCS scores in deep coma.

Main Methods:

• Recorded EEG signals from 39 deep coma patients (GCS 3-8).
• Processed EEG signals into alpha, beta, delta, and theta sub-bands, calculating power spectral density.
• Extracted 10 time and frequency domain features, statistically analyzed, and used machine learning for classification.

Main Results:

• Identified decreased theta activity as a significant differentiator for GCS 3 and GCS 8 patients.
• Achieved 96.44% classification performance in distinguishing patients within deep coma states.
• Demonstrated the potential of EEG-based features for objective LeOC assessment.

Conclusions:

• EEG signal analysis, particularly theta activity, offers an objective method for assessing consciousness levels in deep coma.
• This approach can aid in more precise neurological evaluations and patient stratification.
• The study presents a novel, high-performance method for classifying GCS scores in severe brain injury.