A growing number of studies have demonstrated the role of quantitative
electroencephalography in assessing brain function in neuro-intensive care
units. Still, few studies have examined patients with large hemisphere infarction. Thirty patients with large hemisphere infarction were included
in this preliminary study, and the patients were divided into the death group (twelve
patients) and survival group (eighteen patients). Electroencephalography monitored the patients, and
a computerized tomography inspection was performed. The quantitative electroencephalography of
the alpha-beta/delta-theta ratio change index was calculated and used to
predict the prognosis of early large hemisphere infarction patients. The relationship between three months
modified Rankin Scale, and alpha-beta/delta-theta ratio change
index was analyzed. The death group had negative changes for alpha-beta/delta-theta ratio
change index (-0.0140
Large hemispheric infarction (LHI) is an ischemic stroke affecting the entire or total sub- territory of the middle cerebral artery, involving the basal ganglia at least partially, with or without the adjacent territories’ involvement [1]. It continues to be associated with significant mortality and morbidity, and patients with LHI have a poor prognosis. Therefore, LHI is also known as malignant middle cerebral artery infarction (MCHI) [2], and its incidence accounts for about 5%–10% of all ischemic strokes [3].
As an examination technique of brain bioelectric activity, electroencephalography can directly reflect the abnormal state of brain function and metabolism. It is more sensitive to brain damage caused by the reduction of cerebral blood flow, and its sensitivity is higher than computerized tomography (CT) and other imaging examinations [4]. Quantitative EEG (QEEG) is a quantitative analysis method based on primitive EEG, which is currently widely used in mental diseases, epilepsy, traumatic brain injury, cerebrovascular diseases, Parkinson’s disease, cognitive related, sleep, and many pain diseases. The advantage of QEEG is that it abandons the disadvantage of high requirements of the original EEG on judging readers. The computer generates data, and the result is more objective and reliable. Clinical studies have revealed that QEEG parameters change earlier than clinical indicators, and its predictive value is higher than imaging data and clinical scales [5, 6, 7]. Several studies have shown that higher delta power, lower-alpha power and beta power are reliable QEEG indicators for poor prognosis in patients with acute ischemic stroke (IS) [5, 7, 8, 9]. Studies also indicated that the ratio of delta-theta to alpha-beta (DTABR) and DAR indicators have a high prognostic ability for IS patients [6, 7, 10].
However, studies about QEEG in early LHI are limited. Only two case-report studies revealed that an increase of delta power in the opposite hemisphere suggests a poor prognosis [11], and [12] found that theta power is meaningful in predicting the outcome of death in patients with LHI and is superior to visual EEG and Glasgow Coma Scale (GCS). Alpha-beta/delta-theta ratio (ABDTR) is positively correlated with the patients’ prognosis. The authors in [8] have correlated ischemic cortical stroke patients’ clinical outcomes with acute QEEG, DWI, and PWI data and verified that the 30-day National Institute of Health Stroke Scale (NIHSS) had equivalent correlations with ADCI (acute delta change index) and MTT abnormality volume, both were more significant than the correlation between acute DWI lesion volume and 30-day NIHSSS.
The detailed information of LHI patients confirmed by head CT or MR in the Intensive Care Unit (ICU) of the Second Affiliated Hospital of Hebei Medical University from August 2017 to July 2020 was collected. This study was reviewed and approved by the Ethics Committee of the Second Affiliated Hospital of Hebei Medical University. All patients were managed following Evidence-Based Guidelines for the Management of LHI [1]. This study was registered at the China Clinical Trial Center, the review resolution number was 2017-P018, and the registration number was ChiCTR1800016468.
Collect the patients baseline clinical and diagnostic data, such as age, gender,
risk factors, complication [Lung infection, Acute Gastric Mucosal Lesions, deep
vein thrombosis of the lower extremity, and the cerebral hernia], GCS and NIHSS
on admission, ICU hospital stay (day), imaging indicator [infarct side and
volume, the midline shift of septum pellucidum, the midline shift of pineal
gland], EEG monitoring [the time from the first monitoring to the onset (h), the
time interval between two EEGs (h)], and the mRS of 3 months. A cerebral hernia is
defined as cerebral edema whose clinical manifestations are unequal pupils, head
imaging manifests as a mass effect, and midline displacement
Inclusion criteria: (1) Patients older than 18 years; (2) Patients who received
ICU treatment within three days of onset; (3) CT or MRI confirmed that more than
2/3 area of the middle cerebral artery blood supply area was involved, with or
without infarction of the anterior and posterior cerebral arteries; (4)
The informed consent form was signed by the patient’s family
member or guardian; (5) Patients with reduced levels of consciousness, and the
NIHSS scores were higher than 14 (in patients with an infarction in the
non-dominant hemisphere) or higher than 19 (in patients with an infarction in the
dominant hemisphere). Exclusion criteria: (1) Patients with multiple infarctions
of the cerebellum, brain stem, or bilateral cerebral hemispheres; (2) Patients
who received antiepileptic drugs or sedative drugs within 24 hours before EEG
monitoring, which may affect brain electrical activity; (3) Patient’s pre-onset
mRS score
The patients were monitored by using the digital EEG machine (Model Neusen. U) of Boruikang Technology (Changzhou) Co., Ltd within three days of onset, 2 hours each time, and at least two times for each case; the monitoring was performed in the ICU of the neurology department, and the indoor environment was maintained at a constant temperature and humidity [6]. Since standard clinical EEG arrays are sufficient for post-stroke monitoring and prognosis, high-density electrode arrays are unnecessary. The international 10–20 method was adopted for electrode placement, and the international standard 19-lead EEG was connected, including, Fp1, Fp2, F3, F4, F7, F8, C3, C4, T3, T4, T5, T6, P3, P4, O1, and O2. The silver disk electrode was used as the recording electrode (Cz, A1, A2 points are reference electrodes; Fpz points are ground electrodes). Each lead’s resistance was controlled within 5,000 ohms, the sampling rate was 1000 Hz, and the filtering range was 0.053–500 Hz. Clinicians had no access to the patient’s EEG data during the diagnosis and treatment, and the data was stored in the form of European Data Format (EDF) files.
MATLAB (MATLAB R2015a, The MathWorks, Inc. MA, USA) software and its EEGLAB toolbox were used for quantitative analysis. Two measured EEGs within four days of onset were selected, and the EEG index was calculated. For each monitoring, continuous EEG data with a length of about 30 minutes without motion interference were chosen, and the following steps were performed: (1) Preprocessing: power frequency interference was removed by performing 50 Hz notch filtering; the third-order Butterworth filter was selected as the high-pass filter, the -3dB cutoff frequency was 1 Hz; the 8th order Butterworth filter was selected as the low-pass filter, and the -3dB cutoff frequency was 30 Hz. (2) Adaptive noise reduction: eye electrical interference removal and ICA noise reduction processing were performed using the EEGLAB toolbox method. (3) Each piece of EEG data was cut into 2 s epoch for power spectral density analysis, the Welch method was adapted as an algorithm, window length was 2 s, and overlap was 50%. The ABTDR was calculated, the average value of each epoch’s ratio, the band power Alpha, Beta to band power Delta and Theta, namely (PA + PB)/(PD + PT), the frequency range of each band were shown in Table 1. After calculating the ABDTR of each electrode, the average number of all electrodes was considered the average scalp ABDTR. The ABDTR index was calculated by the second ABDTR of EEG monitoring minus the first ABDTR, divided by the first ABDTR value and the intervals (h) [8]. The ABDTR change index reflects not only the direction but also the degree of EEG changes. In this study, we used ABDTR and its change index to measure brain function.
| EEG parameters | Abbreviation | Frequency range |
| Total power | - | 1–30 Hz |
| Delta power | - | 1–4 Hz |
| Theta power | - | 4–8 Hz |
| Alpha power | - | 8–12.5 Hz |
| Beta power | - | 12.5–30 Hz |
| Alpha-beta/delta-theta | ABDTR | 8–30/1–8 Hz |
CT inspection was performed using the Philips Brilliance CT model, the layer thickness and the layer spacing were both 10 mm, and the measurement software was MedViewer. During the detection, the lesion area was manually outlined layer by layer. The volume was calculated as the sum of the total lesion area multiplied by the layer thickness and the layer spacing. The midline displacement distance was measured at the septum pellucidum level and the pineal gland level, respectively. The head CT results of the patients with the enormous infarct volume during hospitalization were included in the fundamental data analysis.
SPSS 22.0 was used for data analysis. The measurement data
conforming to the normal distribution were expressed as mean
A total of 30 patients (20 males and 10 females) with LHI were included in this
study, with an average age of 67.27
| Projects | Total | Survival group | Death group | P |
| (n = 30, %) | (n = 18, %) | (n = 12, %) | ||
| Basic information | ||||
| Age (year) | 67.27 ± 9.94 | 65.72 ± 10.67 | 69.58 ± 8.65 | 0.310 |
| Gender (Male, %) | 20 (66.67) | 12 (66.67) | 8 (66.67) | 1.000 |
| Risk factors | ||||
| Hypertension | 18 (60.00) | 11 (61.11) | 7 (58.33) | 1.000 |
| Atrial fibrillation | 14 (46.67) | 6 (33.33) | 8 (66.67) | 0.073 |
| Ischemic stroke | 8 (26.67) | 3 (16.67) | 5 (41.67) | 0.210 |
| CHD | 12 (40.00) | 5 (27.78) | 7 (58.33) | 0.196 |
| Diabetes | 7 (23.33) | 3 (16.67) | 4 (33.33) | 0.392 |
| Complication | ||||
| Brain herniation | 14 (46.67) | 6 (33.33) | 8 (66.67) | 0.073 |
| Lung infection | 25 (83.33) | 15 (83.33) | 10 (83.33) | 1.000 |
| AGML | 18 (60.00) | 13 (72.22) | 5 (41.67) | 0.136 |
| DVT | 6 (20.00) | 3 (16.67) | 3 (25.00) | 0.660 |
| GCS on admission | 6.73 ± 3.05 | 8.00 (4.00, 9.25) | 5.83 ± 3.16 | 0.108 |
| NIHSS on admission | 22.52 ± 6.30 | 20.29 ± 4.61 | 25.67 ± 7.10 | 0.056 |
| ICU hospital stay (Day) | 24.00 (21.75, 28.50) | 24 (20.75, 25.00) | 28.92 ± 10.69 | 0.212 |
| Imaging indicator | ||||
| Infarct site (Left) | 18 (60.00) | 13 (72.22) | 7 (58.33) | 0.461 |
| Infarct volume | 400.69 ± 213.73 | 331.77 ± 166.61 | 504.09 ± 241.03 | 0.119 |
| The midline shift of septum pellucidum | 6.00 ± 5.15 | 5.29 ± 4.23 | 7.07 ± 6.34 | 0.051 |
| The midline shift of pineal gland | 4.19 ± 3.79 | 3.86 ± 3.20 | 4.69 ± 4.65 | 0.097 |
| EEG monitoring | ||||
| The time from the first monitoring to the onset (h) | 29.99 (22.36, 53.22) | 29.95 (21.25, 46.58) | 39.39 ± 23.72 | 0.932 |
| The time interval between two EEGs (h) | 20.83 ± 14.42 | 22.22 ± 14.75 | 18.82 ± 14.39 | 0.813 |
| The measurement data conform to the normal distribution and uses the Mean
| ||||
In terms of the change index of ABDTR, a more severe and rapid brain function
decline was noticed in the death group (-0.0140
The ABDTR change index diagnostic results showed that
the area under the curve, sensitivity,
specificity, and the Youden index was 0.815, 75.00%, 83.33%, and 58.33%,
respectively (Z = 3.552, P
Fig. 1.The ROC Curve of the ABDTR change index and Glasgow Coma Scale (GCS).
| Factor | AUC | 95% CI | Sensitivity (%) | Specificity (%) | Youden | Related Standards | Z | P |
| GCS | 0.674 | 0.479–0.833 | 91.67 | 50.00 | 0.417 | ≤8 | 1.675 | 0.094 |
| ABDTR change index | 0.815 | 0.631–0.932 | 75.00 | 83.33 | 0.5833 | ≤-0.008 | 3.552 |
According to the correlation analysis results, the 3-month mRS was significantly negatively correlated with ABDTR change index (r = -0.489, P = 0.006) (Fig. 2).
Fig. 2.The correlation between the ABDTR change index and the 3-month mRS detection.
Typical patient: A 79-years-old male with a previous history of hypertension, diabetes, and coronary heart disease was admitted to our hospital. A specific dose of rtPa (0.9 mg/kg) for thrombolysis was given 46 minutes 2 hours after onset, 5 points for GCS on admission, and 30 points for NIHSS. The head CT at 23 h after the onset was shown in Fig. 3, and continuous long-term monitoring was started 24 h after the onset. Every half hour, a constant 10 minutes of non-interfering EEG data was taken as the characteristics of the half-hour period. The ABDTR was calculated, and the line chart was generated as above. Thirty-nine hours after the onset (the 15 h point), the pupils were unequal in size. The patient’s right pupil diameter was 6.0 mm, and the light reflection disappeared; the left pupil diameter was 3 mm, and the light reflection was dull, and his vital signs were stable. About 4 hours later, the head CT was re-checked and was shown in the figure. Because the patient went out for a head CT examination, the EEG monitoring was stopped. The total duration of the EEG monitoring was 18.5 hours. It can be seen that the patient’s ABDTR has continued to decline before the clinical change (Fig. 4).
Fig. 3.The CT results of typical cases. (A) Onset 23 h CT scan. (B) Onset 43 h CT scan. The red line is the lesion.
Fig. 4.The typical cases of ABDTR result in continuous long-term monitoring after 24 hours of onset. Down arrow: the time when the pupil size of both eyes was not equal.
This study showed that the ABDTR change index was significantly different
between the death and survival groups (P = 0.004). Using the ABDTR
change index as a diagnostic indicator, the AUC, sensitivity and specificity were
0.815, 75.00%, 83.33%, respectively. The results illustrated that the ABDTR change
index has a higher predicted diagnostic value than the value of GCS.
Simultaneously, there was also a significant correlation between the ABDTR change
index and the 3-month mRS score (P
EEG is mainly caused by excitatory and inhibitory postsynaptic currents in the dendrites of cortical pyramidal cells and is very sensitive to detect cerebral ischemia. Ischemic stroke can have various abnormal typical manifestations on an electroencephalogram. Whole-brain delta (absolute) power often changes significantly over time, depending on the severity and evolution of stroke [8]. At moderate ischemic levels, before the delta frequency appears, beta (usually 14–30 Hz) attenuation may also occur [13], followed by slower alpha (typically 8–13 Hz) activity [14].
Some evidence from joint EEG and neuroimaging observations and animal studies indicate that alpha activity disorders (e.g., amplitude attenuation, slowing) usually indicate cortical damage, while abnormal delta frequencies reflect subcortical and/or white matter damage the cortex goes in [15, 16]. However, this is just a conjecture, and we need more evidence to assess these possibilities finally. Of the 20 QEEG measurements calculated using Fourier transform or other analysis, the delta power measurement has the strongest negative correlation with regional cerebral blood flow (rCBF). Other evidence suggests that the delta frequency may reflect pathophysiological processes such as oxidative stress [17].
There are very few quantitative EEG studies on LHI. Finnigan SP et al. [11] found that the contralateral hemisphere delta power can be a predictor of poor prognosis. The authors in [12] found that the contralateral hemisphere’s theta power can predict poor prognosis, but the diagnostic value is moderate. They speculated that the mix of delta activity and artifacts resembling delta activity eliminated the significant difference in delta power and affected the analysis of DTABR, which also involves delta power. Therefore, they suggested that more attention should be paid to theta power in the quantitative EEG study of LHI. A study suggests that theta power may be related to patients with cerebral edema.
The authors in [9] found that theta power is the best predictor of short-term results (along with alpha power) because it is related to dysfunctional but still viable tissues, which may be related to the ischemic penumbra or edema area. In this study, we believe that theta power detection has definite clinical significance in LHI patients. LHI has a larger infarct core and a larger edema volume compared with patients with mild cerebral infarction.
According to previous studies, we think that the power of delta, theta, beta, and alpha may all have significance in developing LHI disease. Therefore, the ABDTR change index was selected, and it includes multiple power indicators that may affect the prognosis, which can improve the diagnostic value. As a prospective study, this study minimizes various artifacts during EEG monitoring implementation, ensuring the quality of EEG monitoring and high diagnostic value.
Other research results suggest that theta and beta frequencies are unreliable in pathophysiology after stroke as delta and alpha frequencies [5, 10, 13, 18, 19]. It may be because these studies excluded cases with a GCS of fewer than 6 points, and the included population had a small infarct size and did not produce large-scale cerebral edema. Therefore, the results of these trials do not apply to the diagnosis and research of LHI.
In early LHI patients, especially within three days, the predictive value of various parameters is not high, such as imaging indicators, multiple scores, and original EEG monitoring. There may be different clinical outcomes in the same infarct volume because of the influence of numerous factors, such as age, degree of brain atrophy, and systemic stress response. EEG monitoring is sensitive to time changes. We used time parameters and the ABDTR change index, obtained from two EEG monitoring indicators to evaluate the prognosis. This indicator can reflect the degree of brain function change and the direction of change. Many studies have proved the prognostic significance of GCS, NIHSS, and infarct volume for LHI [20, 21, 22, 23]. Our study showed no statistically significant difference between GCS and NIHSS in the survival and death groups. The ABDTR change index is statistically different between the two groups, and the diagnostic effect of this index is also better than that of GCS and NIHSS. Therefore, we believe that the ABDTR change index has good value in the LHI forecasting process. The NIHSS score is limited in patients with aphasia and coma, so this article did not use the NIHSS score but used the GCS score. We demonstrated that the ABDTR change index is more valuable than GCS diagnosing of death through our experiment.
Computer-aided EEG interpretation systems have shown the potential to enhance non-expert recognition of many EEG abnormalities in adult neurologically critical patients [24]. Therefore, the extraction of specific EEG parameters for specific diseases will provide great help for the clinical monitoring of nerve functions at the bedside to detect changes in the patient’s condition and guide clinical treatment. We found that ABDTR may also become a valuable indicator in long-range EEG monitoring by presenting a typical case.
Our findings are based on the small size of the population. There are twofold reasons for the small sample size: First, there are relatively few patients with LHI; Second, this is preliminary research with limited experimental personnel and experimental time. Another limitation is that the interval between two EEGs was not uniform. However, we improved through the test method. We have used dividing by the time interval to minimize the interference caused by different time intervals. Therefore, multi-center clinical trials are required in the future to verify the results of this trial.
This experiment demonstrates that the ABDTR change index may be a valuable indicator for predicting the prognosis of LHI. A multi-center clinical trial is required in the future to verify the results of this trial.
ABDTR, alpha-beta/delta-theta ratio; ADCI, acute delta change index; CHD, coronary heart disease; DTABR, ratio of delta-theta to alpha-beta; EDF, European Data Format; GCS, Glasgow Coma Scale; ICU, Intensive Care Unit; IS, ischemic stroke; LHI, large hemispheric infarction; MCHI, middle cerebral artery infarction; mRS, modified Rankin Scale; NIHSS, National Institute of Health Stroke Scale; QEEG, quantitative electroencephalography; rCBF, regional cerebral blood flow.
JT, LG and LDL conceived and designed the study. YZ and ZZ performed data acquisition. YHP and DCL analyzed the data. JT and LG wrote the paper. All authors approved the final article.
This study was obtained with the informed consent of all participants. The institutional review board of the Ethics Committee of the Second Affiliated Hospital of Hebei Medical University approved this study, code 2017-P018.
We thank two anonymous reviewers for their excellent criticism of the article.
This work was supported by the Youth Science and Technology Project of Hebei (grant number 20190529).
The authors declare no conflict of interest.