Detecting awareness in patients with disorders of consciousness using a hybrid brain–computer interface

Yanbin He,Ronghao Yu,Jiahui Pan,Yuanqing Li,Fei Wang,Qiuyou Xie,Haibo Di,Steven Laureys

doi:10.1088/1741-2560/11/5/056007

Abstract

Objective. The bedside detection of potential awareness in patients with disorders of consciousness (DOC) currently relies only on behavioral observations and tests; however, the misdiagnosis rates in this patient group are historically relatively high. In this study, we proposed a visual hybrid brain–computer interface (BCI) combining P300 and steady-state evoked potential (SSVEP) responses to detect awareness in severely brain injured patients. Approach. Four healthy subjects, seven DOC patients who were in a vegetative state (VS, n = 4) or minimally conscious state (MCS, n = 3), and one locked-in syndrome (LIS) patient attempted a command-following experiment. In each experimental trial, two photos were presented to each patient; one was the patientʼs own photo, and the other photo was unfamiliar. The patients were instructed to focus on their own or the unfamiliar photos. The BCI system determined which photo the patient focused on with both P300 and SSVEP detections. Main results. Four healthy subjects, one of the 4 VS, one of the 3 MCS, and the LIS patient were able to selectively attend to their own or the unfamiliar photos (classification accuracy, 66–100%). Two additional patients (one VS and one MCS) failed to attend the unfamiliar photo (50–52%) but achieved significant accuracies for their own photo (64–68%). All other patients failed to show any significant response to commands (46–55%). Significance. Through the hybrid BCI system, command following was detected in four healthy subjects, two of 7 DOC patients, and one LIS patient. We suggest that the hybrid BCI system could be used as a supportive bedside tool to detect awareness in patients with DOC.

Highlights

Patients suffering from a severe brain injury may fall into a coma and develop a variety of different clinical awareness states
We demonstrated that the performance of idle state detection was better for the hybrid brain–computer interface (BCI) than for the P300- or steady-state evoked potential (SSVEP)-based BCI
Considering the above factors, we proposed a visual hybrid BCI combining P300 and SSVEP, a variant of our previously established system (Li et al 2013), for the detection of awareness in disorders of consciousness (DOC) patients in this study

Summary

Introduction

Patients suffering from a severe brain injury may fall into a coma and develop a variety of different clinical awareness states. Functional magnetic resonance imaging (fMRI) and electroencephalographic (EEG) techniques have been proposed to examine the residual brain function in some DOC patients (see Chatelle et al 2012, Liberati and Birbaumer 2012, Noirhomme et al 2013 for review) These studies aimed to detect command-specific changes in fMRI or EEG signals and to provide motor-independent evidence of awareness. Eight patients (four VS, three MCS, and one LIS patient) and four healthy subjects participated in our experiment; three patients (one VS, one MCS and one LIS patient) were able to follow commands using our hybrid BCI (classification accuracy 70–78%) These results implied that the three patients possessed residual cognitive function and conscious awareness, which were detected by our BCI system.

Subjects

Data processing

Results

Discussions