Table 1.
Summary of studies using resting-state EEG for diagnosis, prognosis, and evaluation of intervention and basic researches.
| Objectives | Literatures | Methods | Subjects | Accuracy/sensitivity/specificity (%) | Main results |
|---|---|---|---|---|---|
| Diagnosis | Coleman et al. (24) | Spectrum power ratio | MCS 4, VS 6 | –/–/– | VS showed significantly higher EEG power ratio than MCS |
| Schnakers et al. (25) | BIS | VS 32, Coma 11 | –/75/75 | BIS could differentiate unconscious from conscious | |
| Schnakers et al. (26) | EMCS 13, MCS 30, VS 13, Coma 16 | –/–/– | |||
| Pollonini et al. (27) | Coherence, Granger causality | MCS 7, SND 9 | 100/–/– | Number of connections within and between brain regions could differentiate MCS from SND | |
| Sara and Pistoia (28) | ApEn | VS 10, control 10 | –/–/– | ApEn was lower in VS than in controls | |
| Sarà et al. (29) | VS 38, control 40 | –/100/97.5 | |||
| Wu et al. (30) | Lempel–Ziv complexity, ApEn, cross-approximate entropy | MCS 16, VS 21, control 30 | –/–/– | VS had lowest non-linear indices than MCS and control had highest indices | |
| Gosseries et al. (31) | State entropy, response entropy | MCS 26, VS 24, Coma 6 | –/89/90 | EEG entropy of MCS was higher than VS | |
| Wu et al. (32) | Cross-approximate entropy | MCS 20, VS 30, control 30 | –/–/– | Interconnection of local and distant cortical networks in MCS was superior to that of VS | |
| Landsness et al. (33) | Slow wave activity | MCS 6, VS 5 | –/–/– | MCS showed an alternating sleep pattern;VS preserved behavioral sleep but no sleep EEG patterns; | |
| Leon-Carrion et al. (34) | Coherence, Granger causality | MCS 7, SND 9 | –/–/– | MCS showed frontal cortex disconnection from other cortical regions | |
| Significant difference in full bandwidth coherence between SND and MCS | |||||
| Lehembre et al. (35) | Spectrum power, coherence, imaginary part of coherence, phase lag index | MCS 18, VS 10, Acute/subacute 15 | –/–/– | VS showed increased delta, decreased alpha power, and lower connectivity than MCS | |
| King et al. (36) | wSMI | MCS 68, VS 75, CS 24, control 14 | –/–/– | wSMI increases as a function separate VS from MCS | |
| Malinowska et al. (37) | Matching pursuit decomposition, Slow wave activity, K-complexes | LIS 1, MCS 20, VS 11 | 87/–/– | Sleep EEG patterns correlated with patients’ diagnosis | |
| Bonfiglio et al. (38) | Blink-related delta oscillations | MCS 5, VS 4, control 12 | –/–/– | Patients showed abnormal blink-related delta oscillations | |
| Lechinger et al. (39) | Spectrum power | MCS 9, VS 8, control 14 | –/–/– | Ratios between frequencies (above 8 Hz) and (below 8 Hz) correlated with CRS-R | |
| Höller et al. (40) | A total of 44 indices | MCS 22, VS 27, control 23 | Partial coherence: MCS vs. VS (88), control vs. MCS (96), control vs. VS (98) | Connectivity was crucial for determining the level of consciousness | |
| Transfer function: MCS vs. VS (80), control vs. MCS (87), control vs. VS (84) | |||||
| Partial coherence: MCS vs. VS (78), control vs. MCS (93), control vs. VS (96) | |||||
| Sitt et al. (18) | Spectrum power, spectral entropy, Kolmogorov–Chaitin complexity, phase locking index, wSMI, permutation entropy | MCS 68, VS 75, CS 24, control 14 | Best cross-validated single measure: MCS vs. VS (AUC = 71 ± 4) | ||
| Whole set of measures: MCS vs. VS (AUC = 78 ± 4) | |||||
| The most discriminative measure was wSMI, which separated VS from MCS | |||||
| Marinazzo et al. (41) | Multivariate Granger causality, transfer entropy | MCS 10, EMCS 5, VS11, control 10 | –/–/– | In VS, the central, temporal, and occipital electrodes showed asymmetry between incoming and outgoing information | |
| Bonfiglio et al. (42) | Blink-related synchronization/desynchronization | MCS 4, VS 5, control 12 | –/–/– | Blink-related synchronization/desynchronization could differentiate MCS from VS | |
| Naro et al. (43) | Spectrum power, LORETA | MCS 7, VS 6, control 10 | –/–/– | Alpha was the most significant LORETA data correlating with the consciousness level | |
| Piarulli et al. (44) | Spectrum power, spectral entropy | MCS 6, VS 6 | –/–/– | MCS showed higher theta and alpha, lower delta, higher spectral entropy, and higher time variability than VS | |
| Thul et al. (45) | Permutation entropy, symbolic transfer entropy | MCS 7, VS 8, control 24 | Permutation entropy: Control vs. MCS (Max AUC = 0.74), control vs. VS (Max AUC = 0.91), MCS vs. VS (Max AUC = 0.74) | ||
| Symbolic transfer entropy: Control vs. MCS (Max AUC = 0.80), control vs. VS (Max AUC = 0.80), MCS vs. VS (Max AUC = 0.71) | |||||
| Chennu et al. (46) | dwPLI, brain network | MCS 66, VS 23, control 26 | VS vs. MCS: Alpha participation coefficient (AUC = 0.83, accuracy = 79%), alpha median connectivity (AUC = 0.82), alpha modular span (AUC = 0.78) | ||
| MCS− vs. MCS+: delta power averaged over all channels (AUC = 0.79) | |||||
| Prognosis | Babiloni et al. (47) | Cortical sources estimated by LORETA | VS 50, control 30 | Power of alpha source predicted the follow-up recovery | |
| Wu et al. (30) | Lempel–Ziv complexity, ApEn, cross-approximate entropy | MCS 16, VS 21, control 30 | Non-linear indices of patients who recovered increased than those in non-recovery | ||
| Fingelkurts et al. (48) | EEG oscillatory microstates | MCS 11, VS 14 | Diversity and variability of EEG for non-survivors were significantly lower than for survivors | ||
| Sarà et al. (29) | ApEn | VS 38, control 40 | Patients with lowest ApEn either died or remained in VS, patients with highest ApEn became MCS or partial or full recovery | ||
| Cologan et al. (49) | Sleep spindles | MCS 10, VS 10 | Patients who clinically improved within 6 months have more sleep spindles | ||
| Arnaldi et al. (50) | Sleep patterns | MCS 6, VS 20 | Sleep patterns were valuable predictors of a positive clinical outcome in sub-acute patients | ||
| Schorr et al. (51) | Spectrum power, coherence | MCS 15, VS 58, control 24 | Short- and long-range coherence had a diagnostic value in the prognosis of recovery from VS | ||
| Wislowska et al. (52) | Spectral power, sleep patterns, permutation entropy | MCS 17, VS 18, control 26 | Sleep patterns did not systematically vary between day and night in patients | ||
| Day–night changes in EEG power spectra and signal complexity were revealed in MCS, but not VS | |||||
| Sleep patterns were linearly related to outcome | |||||
| Chennu et al. (46) | dwPLI, brain network | MCS 66, VS 23, control 26 | Delta band connectivity and network had a clear relationship with outcomes | ||
| Treatment evaluation | Williams et al. (53) | Spectrum power, coherence, zolpidem | Patients response in zolpidem 3 | Spectral peak of 6–10 Hz with high spatial coherence was a predictor of zolpidem responsiveness | |
| Manganotti et al. (54) | Spectrum power, 20 Hz rTMS | MCS 3, VS 3 | rTMS over M1 induced long-lasting behavioral and neurophysiological modifications in one MCS patient | ||
| Carboncini et al. (55) | Spectrum power, phase synchronization, midazolam | MCS 1 | Change in the power spectrum was observed after midazolam | ||
| Midazolam induced significant connectivity changes | |||||
| Cavinato et al. (56) | Coherence, simple sensory stimuli | MCS 11, VS 15 | Increase in short-range parietal and long-range fronto-parietal coherences in gamma frequencies was seen in the controls and MCS | ||
| VS showed no modifications in EEG patterns after stimulation | |||||
| Pisani et al. (57) | Slow wave activity, 5 Hz rTMS | MCS 4, VS 6 | Following the real rTMS, a preserved sleep–wake cycle, a standard temporal progression of sleep stages appeared in all MCS but none of VS | ||
| Naro et al. (58) | Spectrum power, coherence, tACS | MCS 12, VS 14, control 15 | TACS entrained theta and gamma oscillations and strengthened the connectivity patterns within frontoparietal networks in all the control, partial MCS, and some VS | ||
| Naro et al. (59) | Spectrum power, coherence, otDCS | MCS 10, VS 10, control 10 | Fronto-parietal networks modulation, theta and gamma power modulation, and coherence increase were paralleled by a transient CRS-R improvement, only in MCS individuals | ||
| Naro et al. (60) | Lagged-phase synchronization, network parameters, rTMS | MCS 9, VS 11, control 10 | Two VS patients showed a residual rTMS-induced modulation of the functional correlations between the default mode network and the external awareness networks, as observed in MCS | ||
| Bai et al. (61) | Relative power, coherence, biocoherence, SCS with 5, 20, 50, 70, 100 Hz | MCS 11 | Significantly altered relative power and synchronization was found in delta and gamma bands after one SCS stimulation using 5, 70, or 100 Hz | ||
| Bicoherence showed that coupling within delta was significantly decreased after stimulation using 70 Hz | |||||
| Basic research | Davey et al. (62) | Spectrum power, coherence | VS 1 | Greater low-frequency power, less high-frequency power, and reduced coherence were over the more damaged right hemisphere | |
| Babiloni et al. (63) | Spectrum power, LORETA | LIS 13, control 15 | Power of delta and alpha was abnormal in LIS | ||
| King et al. (36) | EEG oscillatory microstates | MCS 7, VS 14 | Decreased number of EEG microstate types was associated with altered states of consciousness | ||
| Unawareness was associated with the lack of diversity in EEG alpha-rhythmic microstates | |||||
| Sitt et al. (18) | Spectrum power | LIS 1, MCS 2, control 5 | One MCS and one LIS showed motor imagery task performance through spectral change which was different from control | ||
| Varotto et al. (64) | Partial directed coherence | VS 18, control 10 | VS patients showed a significant and widespread decrease in delta band connectivity, whereas the alpha activity was hyper-connected in the central and posterior cortical regions | ||
| Chennu et al. (46) | dwPLI, graph theoretic network | MCS 19, VS 13, control 26 | Network of patients had reduced local and global efficiency, and fewer hubs in the alpha band | ||
| Forgacs et al. (65) | Sleep patterns | EMCS 13, MCS 23, VS 8 | Patients with evidence of covert command-following had well-organized EEG background and relative preservation of cortical metabolic activity | ||
| Pavlov et al. (66) | VS 15 | Most of VS patients had abnormal sleep patterns | |||
ApEn, approximate entropy; wSMI, weighted symbolic mutual information; tACS, transcranial alternating current stimulation; otDCS, oscillatory transcranial direct current stimulation; rTMS, repetitive transcranial magnetic stimulation; SCS, spinal cortical stimulation; CS, conscious patients; SND, severe neurocognitive disorders; LIS, locked-in syndrome; MCS, minimally conscious state; EMCS, emergence from MCS; VS, vegetative state; dwPLI, debiased weighted phase lag index; AUC, area under the receiver operating characteristic curve.