Emotion-Related Consciousness Detection in Patients With Disorders of Consciousness Through an EEG-Based BCI System

Jiahui Pan, Qiuyou Xie, Haiyun Huang, Yanbin He, Yuping Sun, Ronghao Yu, Yuanqing LiFrontiers in Human Neuroscience 2018 · DOI

Abstract

Detecting residual cognitive function in patients with disorders of consciousness (DOC), such as vegetative state (VS) and minimally conscious state (MCS), is a significant clinical challenge. Traditional assessments like the Coma Recovery Scale-Revised (CRS-R) rely on motor responses, which many DOC patients cannot provide due to severe physical impairments. This study presents an EEG-based brain-computer interface (BCI) designed to detect consciousness through emotion recognition. By presenting emotional movie clips and analyzing P300 potentials and spectral features, the system identified command following and emotional processing in several patients who appeared behaviorally unresponsive.

Introduction

Patients in a VS or MCS often possess "silent" cognitive abilities that are masked by severe motor impairments. Misdiagnosis rates remain high because clinical gold standards depend on overt behavioral responses. Emotion recognition, a high-level cognitive process involving broad neural networks, offers a potential alternative marker for awareness. This research explores whether an EEG-based BCI can bypass motor requirements to detect hidden consciousness by observing neural responses to "happy" and "sad" stimuli.

Methods and Experimental Design

The study involved eight DOC patients and eight healthy controls. Participants were presented with two movie clips—one depicting laughing and the other crying—which flashed periodically to evoke event-related potentials (ERPs).

Experimental interface showing the laughing and crying movie clips used as stimuli to evoke P300 potentials.

The BCI system utilized a two-pronged approach:

  1. Online P300 Detection: Subjects were instructed to focus on and count the flashes of a specific target clip. A Support Vector Machine (SVM) was used to detect the P300 potential, a marker of selective attention and stimulus recognition, to determine which clip the patient was attending to.
  2. Offline Spectral Analysis: EEG signals were analyzed across five frequency bands (delta, theta, alpha, beta, and gamma) using Common Spatial Pattern (CSP) filters and Differential Entropy (DE) to classify emotional states.

Key Results

The BCI successfully detected signs of consciousness in a subset of the patient cohort. Three out of eight DOC patients (37.5%) achieved online P300 detection accuracies significantly above the chance level (64%), with the highest accuracy reaching 78%.

Grand-average ERP waveforms showing a distinct P300 peak (red solid line) for target stimuli compared to non-target stimuli (blue dashed line) in successful participants.

Offline analysis further supported these findings, with two patients showing significant classification accuracy between happy and sad states based on spectral features. Notably, the topographic distribution of these features in successful patients mirrored those found in healthy controls, suggesting preserved emotional processing pathways. Specifically, sadness was associated with higher delta activity in prefrontal areas, while happiness elicited stronger beta and gamma responses in lateral temporal sites.

Discussion and Prognostic Value

One of the most compelling findings was the prognostic potential of the BCI. A patient initially diagnosed as being in a vegetative state demonstrated significant BCI performance and subsequently progressed to a minimally conscious state one month later. This suggests that neural markers of emotion-related consciousness may precede behavioral recovery. While limited by a small sample size and patient fatigue, this study demonstrates that BCI systems can provide a vital window into the minds of unresponsive patients, offering a more nuanced assessment of their cognitive state than traditional bedside evaluations.

Summary updated

Abstract

For patients with disorders of consciousness (DOC), such as vegetative state (VS) and minimally conscious state (MCS), detecting and assessing the residual cognitive functions of the brain remain challenging. Emotion-related cognitive functions are difficult to detect in patients with DOC using motor response-based clinical assessment scales such as the Coma Recovery Scale-Revised (CRS-R) because DOC patients have motor impairments and are unable to provide sufficient motor responses for emotion-related communication. In this study, we proposed an EEG-based brain-computer interface (BCI) system for emotion recognition in patients with DOC. Eight patients with DOC (5 VS and 3 MCS) and eight healthy controls participated in the BCI-based experiment. During the experiment, two movie clips flashed (appearing and disappearing) eight times with a random interstimulus interval between flashes to evoke P300 potentials. The subjects were instructed to focus on the crying or laughing movie clip and to count the flashes of the corresponding movie clip cued by instruction. The BCI system performed online P300 detection to determine which movie clip the patients responsed to and presented the result as feedback. Three of the eight patients and all eight healthy controls achieved online accuracies based on P300 detection that were significantly greater than chance level. P300 potentials were observed in the EEG signals from the three patients. These results indicated the three patients had abilities of emotion recognition and command following. Through spectral analysis, common spatial pattern (CSP) and differential entropy (DE) features in the delta, theta, alpha, beta, and gamma frequency bands were employed to classify the EEG signals during the crying and laughing movie clips. Two patients and all eight healthy controls achieved offline accuracies significantly greater than chance levels in the spectral analysis. Furthermore, stable topographic distribution patterns of CSP and DE features were observed in both the healthy subjects and these two patients. Our results suggest that cognitive experiments may be conducted using BCI systems in patients with DOC despite the inability of such patients to provide sufficient behavioral responses.
BibTeX
@inproceedings{pan2018emotion,
  author = {Jiahui Pan and Qiuyou Xie and Haiyun Huang and Yanbin He and Yuping Sun and Ronghao Yu and Yuanqing Li},
  title = {{Emotion-Related Consciousness Detection in Patients With Disorders of Consciousness Through an EEG-Based BCI System}},
  booktitle = {Frontiers in Human Neuroscience},
  year = {2018},
  doi = {10.3389/fnhum.2018.00198},
}
Figure