Visual abstract

Introduction

Central post-stroke pain (CPSP) is a representative central neuropathic pain syndrome caused by lesions involving the thalamus or the spinothalamic tract. It is characterized by various sensory disturbances including spontaneous pain, allodynia, and hyperalgesia [1]. Clinically, such symptoms have been recognized for more than 1,500 years. In Cheon-geum-yo-bang (Qianjin Yaofang in Chinese), written in 652 CE by Son Sa-mak (Sun Simiao in Chinese) of the Dang (Tang in Chinese) Dynasty, “Pung-bi” (“fengbi” in Chinese) was described as 1 of the 4 major manifestations of stroke. Following a stroke, “Pung-bi” encompasses severe limb pain, sensory abnormalities, numbness, and burning sensations features which closely aligned with the contemporary definition of CPSP [2,3].

From a modern medical perspective, CPSP remains a significant clinical issue. Strokes are a leading global cause of death and disability, with approximately 101 million stroke survivors reported in 2019 [4]. Up to 55% of these individuals experience CPSP [5], which can profoundly deteriorate their quality of life. Beyond pain itself, CPSP commonly leads to depression, anxiety, and sleep disturbance, and in severe cases, may even provoke suicidal behavior. CPSP interferes with rehabilitation participation and functional recovery, thereby further worsening long-term outcomes [1].

Despite its clinical importance, the diagnosis and assessment of CPSP has substantial limitations. Commonly used scales such as the Modified Barthel Index and the Fugl-Meyer Assessment primarily focus on motor function, whereas widely used pain scales including the Numerical Rating Scale, Visual Analog Scale, and Neuropathic Pain Symptom Inventory (NPSI) rely heavily on subjective self-report [1]. Given that CPSP presentation varies depending on lesion location, the degree of thalamocortical reorganization, and the patient’s emotional or cognitive status, subjective measures alone cannot adequately capture underlying pathophysiological changes or sensitively monitor treatment responses [1]. This underscores the urgent need for objective, physiology-based assessment tools.

Quantitative electroencephalography (qEEG) has emerged as a promising modality that may fulfill this need. qEEG characterizes electrical brain activity quantitatively and detects pathophysiological alterations noninvasively with millisecond-level temporal resolution. This enables real-time tracking of neural network changes involved in pain perception and modulation. Spectral band power analysis, spectral exponent measures, connectivity indices, and nonlinear metrics hold the potential to objectively reflect key CPSP mechanisms including thalamocortical rhythm disturbances, and disruptions in the excitatory-inhibitory balance [6]. However, qEEG studies specifically dedicated to CPSP remain scarce, and its diagnostic specificity and clinical applicability have not been fully elucidated. Against this background, a study is outlined in this protocol which aims to explore the potential role and clinical utility of qEEG in the objective assessment of CPSP.

Materials and Methods

1. Study registration and ethical approval

This study protocol was registered with the Clinical Research Information Service of the Korea National Institute of Health, Republic of Korea (KCT0011088, date of registration: 30 October 2025). The study was reviewed and approved by the Institutional Review Board of Wonkwang University Korean Medicine Hospital in Gwangju (WKIRB 2025/14-2, 17 September 2025). All procedures will be conducted in accordance with the ethical principles of the Declaration of Helsinki.

2. Study design

This study was designed to be a single-center, prospective cross-sectional observational study. Written informed consent will be obtained from all participants, together with demographic information, stroke history, presence and location of central neuropathic pain, responsiveness to cold stimulation (cold allodynia), major medical conditions within the past 5 years including prior surgeries, current medication use, and questionnaire data will be collected including the NPSI and the Patient Health Questionnaire-9 (PHQ-9).

3. Eligibility

3.1. Target conditions and symptoms

Patients with central neuropathic pain following stroke will be assigned to the CPSP group and those patients without pain after stroke will be in the non-CPSP group.

3.2. Inclusion criteria

The inclusion criteria: (1) Adults aged ≥ 19 years; (2) Diagnosed with stroke (ischemic stroke, hemorrhagic stroke, or subarachnoid hemorrhage) based on clinical presentation and neuroimaging findings (International Classification of Diseases-10, codes I60–I63); (3) At least 3 months (90 days) since the onset of stroke; (4) Alert, able to communicate, and without impairment of expressive language, thus, allowing EEG measurement; and (5) Able to understand the study purpose and voluntarily provide informed consent.

CPSP group inclusion criteria: (1) Development of new pain after the stroke event, characterized as persistent, burning or tingling in nature, and sometimes provoked by cold stimuli; (2) Pain distribution must be neuroanatomically consistent with the lesion identified on neuroimaging; and (3) NPSI total score ≥ 20.

The threshold of an NPSI total score ≥ 20 out of 100 was selected based on prior literature investigating CPSP [7]. In a controlled symptom-psychophysical study by Barbosa et al [7], an NPSI total score ≥ 20 was associated with a high sensitivity (87%) for identifying CPSP, although specificity was relatively low (28%). Based on these findings, the protocol will adopt this threshold for the study to enhance the identification of participants with a clinically meaningful neuropathic pain burden whilst acknowledging that the NPSI is primarily a symptom severity scale rather than a standalone diagnostic instrument. The cut-off was therefore applied as an operational criterion to enrich the CPSP group with individuals exhibiting prominent neuropathic pain characteristics, thereby facilitating group differentiation for qEEG analysis.

The non-CPSP group inclusion criteria: (1) Meets all common inclusion criteria but does not fulfill the CPSP-group criteria.

3.3. Exclusion criteria

Exclusion criteria: (1) Presence of chronic pain prior to the onset of stroke; and (2) Presence of alternative conditions that may cause pain including: complex regional pain syndrome; post-stroke shoulder pain; peripheral neuropathy; fibromyalgia; migraine; and other musculoskeletal disorders. Supplementary criteria suggesting alternative pain etiologies: pain aggravated or relieved by specific movements or postures (e.g., shoulder pain worsened by joint motion, knee osteoarthritis pain); pain directly associated with inflammation or injury of local joints, muscles, or soft tissues (e.g., bursitis, tendinopathy, post-traumatic pain); pain showing a dermatomal (somatic segmental) distribution (e.g., lumbar radiculopathy, sciatica); typical features of peripheral neuropathic pain (e.g., stocking-glove sensory loss, diabetic neuropathy) and paroxysmal headache disorders (e.g., migraine, cluster headache). (3) Neurological diseases other than stroke, such as Parkinson’s disease, dementia, multiple sclerosis, or epilepsy; (4) Severe psychiatric disorders including major depressive disorder, schizophrenia, or anxiety disorders; and (5) Any individual deemed unsuitable for participation at the discretion of the investigator.

4. Recruitment and allocation

The investigators will identify potential participants amongst stroke patients who visit the study site (Wonkwang University Gwangju Korean Medicine Hospital) and will explain the study procedures to assess their willingness to participate. Recruitment posters will also be displayed within and outside the institution to facilitate participant enrollment. Individuals who provide written informed consent will be assigned a participant identification number. A screening number will be issued in the order in which consent is obtained at the initial visit, whereas an enrollment number will be assigned sequentially to those who meet all inclusion and exclusion criteria.

This study does not involve randomization. Participants will be classified into either the CPSP group or the non-CPSP group based on the predefined eligibility criteria regarding the presence or absence of central post-stroke neuropathic pain. No artificial or investigator-driven group allocation will be performed.

5. Quantitative EEG measurement

5.1. qEEG recording device

This study will utilize the Cerowave EEG device (Product name: Cerowave; Model: NGD-01) and its associated software BrainBay (Version 2.1), both manufactured by NeuroGrin Co., Ltd., Seoul, Republic of Korea (Figure 1). Electrode placement follows the International 10–20 system, and EEG signals are recorded from 2 frontal electrodes positioned at Fp1 and Fp2. Dry-type electrodes integrated into the headband are used to secure the sensors to the participant’s forehead. In addition, a photoplethysmography-based reference electrode (attached to the earlobe) will be used to record heart rate simultaneously with EEG acquisition. All signals are sampled at 250 Hz and extracted using a band-pass filter ranging from 1 to 45 Hz.

Although scalp EEG signals recorded from Fp1 and Fp2 primarily reflect cortical surface activity, analysis of phase and amplitude relationships in frontal EEG signals can provide indirect insights into functional alterations within broader pain-related networks including thalamocortical circuits. Frontal EEG measures have been shown to capture network-level oscillatory dynamics associated with pain perception and modulation [8].

The decision to focus on frontal electrodes was also guided by practical considerations. Multichannel EEG acquisition can be time-consuming and technically demanding in clinical settings. Therefore, an analytic framework based on frontal EEG channels enhances feasibility and scalability for potential clinical implementation. Similar simplified approaches have been successfully adopted in other clinical prediction models, such as single-lead electrocardiography for cardiac event prediction, underscoring the importance of balancing neurophysiological depth with clinical applicability [9].

5.2. qEEG recording procedure

To minimize environmental influences on EEG signals, all recordings will be conducted in a quiet clinical room with access restricted to only the investigator and participant. As a stabilization procedure, prior to EEG measurement, participants undergo approximately 3 minutes of passive bilateral eye movements using a commercially available Eye Movement Desensitization and Reprocessing device. This is intended to reduce emotional tension and physiological arousal, thereby enhancing the stability of the EEG baseline [10].

For the CPSP group, EEG measurements will be performed at the time of day when pain symptoms are most prominent. If pain is predominantly nocturnal and measurement during nighttime is impractical, the recording will be conducted at a time when the participant experiences a similar level of pain during normal wakefulness. Sensory characteristics known to provoke or worsen pain will be documented in advance based on patient interview.

Participants will sit comfortably in a chair, and the EEG device will be fitted to ensure secure electrode contact with the frontal region. After confirming that the device is being worn without discomfort, the investigator will power on the device and wait approximately 1 minute for stabilization. The participant will be instructed to fixate on a white cross on a black background, whilst a 5-minute eyes-open (EO) resting-state EEG is recorded. This will be followed by a 5-minute eyes-closed (EC) resting-state EEG. Subsequently, a 2-minute cold-stimulation EEG will be obtained whilst applying an ice pack as the cold stimulus.

Cold stimulation will be performed using a commercially available ice pack (12 × 15 cm) stored in a standard freezer (typically maintained below 0°C, and the ice pack frozen for at least 24 hours prior to application). The ice pack will be wrapped in a thin towel to prevent direct skin injury. In both the CPSP and non-CPSP groups, the ice pack will be applied to the affected side corresponding to the stroke lesion. In the CPSP group, stimulation will be delivered to the area where pain is reported to be most severe. In the non-CPSP group, the ice pack will be applied to the palmar surface of the affected hand. The application site will be kept consistent throughout the stimulation period. The ice pack will be gently placed in contact with the skin without exerting additional pressure. Each stimulation will consist of 30 seconds of application followed by a 10-second rest interval, repeated 3 times. EEG signals will be continuously recorded throughout the entire stimulation protocol.

If the participant produces noise-inducing behaviors such as talking during EO measurement or opening their eyes during EC measurement the investigator will immediately intervene to minimize motion-related or behavioral artifacts. After EC recording has been completed, the device will be powered off. The full process is illustrated in Figure 2.

5.3. qEEG data preprocessing

EEG data extracted from the device in American Standard Code for Information Interchange format will be de-identified prior to analysis. All direct personal identifiers are removed, and only subject ID numbers are retained. The de-identified data will then be transmitted to NeuroGrin Co., Ltd., where signal processing and analysis will be conducted collaboratively with technical experts. NeuroGrin personnel do not have access to identifiable participant information and will be permitted to handle only de-identified datasets. All personnel involved in data handling will sign a confidentiality agreement provided by the Korea Health Technology Research and Development Program, and access privileges will be granted exclusively to those who have completed this process.

During preprocessing, artifact removal will be performed using BrainBay, an open-source software for real-time EEG and biosignal recording. Noise related to electrical interference, muscle contractions, and other myogenic artifacts will be removed. EEG signals will be subsequently transformed via fast Fourier transform to compute spectral power across frequency bands. Both absolute power and relative power, defined as the proportion of absolute power in each frequency band relative to total power will be derived and used as dependent variables for between-group comparisons.

NeuroGrin discards all intermediate data generated during preprocessing immediately after transferring the processed dataset to the principal investigator. All related records will be stored under the supervision of the principal investigator, and upon study completion, they will be maintained or disposed of according to institutional archiving and destruction guidelines.

6. Demographic and clinical data, and qEEG measurements

6.1. Demographic data

Age and sex: confirmed using government-issued identification.

6.2. Clinical data

Questionnaires: Korean version of the NPSI [11], PHQ-9 [12]. Stroke history: date of initial stroke onset, date of initial stroke diagnosis, dates of CT or MRI examinations, stroke type (ischemic stroke, intracerebral hemorrhage, subarachnoid hemorrhage), infarct location (anterior circulation infarct, posterior circulation infarct, lacunar infarct), hemorrhage location (cortical, subcortical, brainstem, cerebellar, intraventricular), and surgical history related to stroke. CPSP-related information: presence or absence of CPSP, pain characteristics, pain-aggravating factors, time of day when pain is most severe, location of pain. Past medical and surgical history: any medical conditions or surgeries other than stroke occurring more than 5 years prior to the screening date. Procedural or medication history: any procedures or medication use within the 3 months preceding the Screening Visit.

6.3. qEEG measurements

Asymmetry features of major frequency bands (delta, theta, alpha, beta, gamma) during EO and EC states: (1) Power in each frequency band (delta: 1–4 Hz, theta: 4–8 Hz, alpha: 8–13 Hz, beta: 13–30 Hz, gamma: 30–100 Hz) will be quantified to assess neural activity [8]; (2) Absolute power represents total energy within each frequency band, while relative power represents the proportion of absolute power relative to total power [8]; (3) As only 2 frontal electrodes (Fp1, Fp2) will be used, spatial comparisons will be limited. Therefore, emphasis will be placed on indices reflecting hemispheric differences rather than raw band power values; (4) Asymmetry feature: A frontal hemispheric symmetry index calculated as the logarithm of the ratio of band power at Fp2 (right frontal) to Fp1 (left frontal) [8]; (5) Asymmetry feature = log (band power at Fp2 / band power at Fp1); (6) The log(Fp2/Fp1) ratio quantifies interhemispheric imbalance in cortical activity and is commonly used to evaluate lateralized frontal activity associated with pain, emotional regulation, and cognitive function [8]; and (7) Specific frequency-band asymmetry has been reported in chronic pain conditions, suggesting its potential as a qEEG biomarker for central neuropathic pain [8].

Delta-to-alpha ratio (DAR) during EO and EC states: (1) DAR represents the ratio of slow-wave delta power to fast-wave alpha power [13]; (2) High DAR values indicate increased slow-wave activity, which are associated with impaired brain function, cognitive slowing, drowsiness, and pain. Low DAR values suggest dominant alpha activity, reflecting normal or alert states with adequate attentional functioning [13]; and (3) Some studies have reported that DAR is more reliable than delta-theta to beta-alpha ratio (DTABR) for evaluating post-stroke physiological and pathological status [13].

DTABR during EO and EC states: (1) DTABR = (Delta Power + Theta Power) / (Alpha Power + Beta Power); and (2) High DTABR indicates dominant slow-wave activity, reflecting reduced cortical function, while low DTABR reflects increased fast-wave activity and a more active neural state. In states such as pain, drowsiness, or cognitive decline, slow-wave activity increases, resulting in elevated DTABR values [14].

6.4. Visit schedule and assessment procedures

The Screening Visit will take place at the participant’s initial visit, the following assessments will be conducted: (1) Obtain written informed consent; (2) Assign a screening number; (3) Collect demographic information; (4) Review stroke history and any past medical history, surgical history, procedures, or medication use unrelated to stroke; (5) Assess eligibility based on inclusion and exclusion criteria; (6) Administer NPSI and PHQ-9 questionnaires; and (7) Schedule the next visit (Visit 1 may be conducted on the same day if eligibility is confirmed).

Visit 1 (group allocation and qEEG measurement) will be conducted within 2 weeks of the Screening Visit. Participants will be assigned to either the CPSP group or the non-CPSP group based on inclusion criteria: (1) CPSP group (document pain location, aggravating stimuli, and time of maximal pain intensity); (2) Perform qEEG measurements; and (3) Assess adverse events (AEs): none, worsening of pain, dizziness, anxiety, or discomfort, other. The flow diagram is shown in Figure 3.

7. Safety assessments and monitoring

7.1. Assessment of AEs

The principal investigator will evaluate the severity of all AEs and serious adverse events (SAEs) reported during the study period. These assessments will be based on the clinical judgment of the principal investigator. The severity (intensity) of each AE and SAE documented in the case report form will be graded with reference to the World Health Organization guidelines. For events not explicitly defined in the guideline, severity will be classified according to the following criteria: (1) Grade 1 (mild) indicates AEs that cause temporary or mild discomfort, do not require treatment, and do not significantly interfere with the participant’s normal daily activities; no limitations in physical functioning; (2) Grade 2 (moderate) indicates AEs that cause mild to moderate limitation in daily activities, significantly impairing normal functioning and potentially requiring some assistance from others. Treatment may or may not be required, and recovery typically occurs with or without intervention; and (3) Grade 3 (severe) indicates AEs that cause marked limitation in daily activities, often requiring assistance from others, and may necessitate medical intervention or hospitalization [15].

7.2. Reporting of SAEs

The principal investigator, study staff, and coordinators will educate study participants (or their parents/legally authorized representatives, if applicable) about the possibility of AEs during study participation and instruct them to report any such occurrences promptly. All systemic or clinically relevant symptoms occurring during the study will be recorded in the case report form including type, onset time, severity, management, medications administered, clinical course, and causal relationship with the study intervention. These procedures must adhere to regulatory standards for clinical trials of drugs and medical devices. If any SAEs or serious drug-related reactions occur during the study, the principal investigator will report them to the study sponsor when submitting the study results. The sponsor will then determine whether the study should be continued or terminated. SAEs must be reported promptly to the Institutional Review Board of the study site in accordance with institutional regulations.

8. Sample size

8.1. Target sample size

A total of 50 participants will be enrolled. CPSP group n = 25 participants and non-CPSP group n = 25 participants.

8.2. Rationale for sample size calculation

For the primary outcome, a significance level (alpha) of 5% and a statistical power of 80% (1-beta) were set. However, no previous studies have evaluated the asymmetry feature in qEEG specifically among patients with CPSP. Therefore, a prior study conducted in patients with chronic neuropathic pain was referred to. It used the same Fp1 and Fp2 electrode positions and employed the asymmetry feature as the primary analytic variable [8]. In that study, the correlation coefficient between beta-band asymmetry and pain scores was reported as r = −0.398 (p = 0.027). Converting this correlation value to Cohen’s d yielded an effect size of 0.87. Using this effect size, the minimum required sample size was calculated with Zα/2 = 1.96 and Zβ = 0.84.

9. Statistical analysis

9.1. General principles of statistical analysis

Since this study will be an observational investigation that involves EEG measurement without any intervention or randomization, the intention-to-treat and per-protocol analysis principles will not be applicable. All collected data will be included in the analysis; however, primary analyses will be conducted using the analyzed set, and supplementary analyses may be performed using the complete case set, which includes only participants without missing values for the primary outcome variables. All statistical analyses will employ 2-sided tests with a significance level of 0.05. Demographic and baseline characteristics will be summarized using descriptive statistics (mean ± standard deviation, frequency, and percentage). Between-group comparisons will be conducted using the following statistical tests depending on normality: (1) Continuous variables: Independent samples t test or Mann-Whitney U test; and (2) Categorical variables: Chi-square test or Fisher’s exact test.

If significant differences are observed in baseline characteristics between groups, the relevant variables may be included as covariates in multivariable analyses to adjust for potential confounding effects.

Medication use, particularly central nervous system (CNS)-active agents such as antidepressants, anticonvulsants (including gabapentinoids), benzodiazepines, and other sedative-hypnotics, will be systematically recorded at baseline, as these agents may influence EEG spectral characteristics. Medication exposure will be categorized into prespecified classes and coded as binary variables (present/absent) for analysis. Given the relatively small sample size, covariate adjustment will be performed parsimoniously to minimize model overfitting.

The primary qEEG outcome of the proposed study will be the beta-band frontal asymmetry feature measured during the EC resting-state condition. This variable was selected based on prior evidence suggesting its association with neuropathic pain severity and altered cortical excitability and was also used as the basis for sample size estimation [6,8].

The statistical hierarchy of outcomes is prespecified as follows: the beta-band frontal asymmetry measured during the EC resting-state condition constitutes the sole primary outcome and will not be subject to multiple-comparison correction. All other qEEG indices including asymmetry features in other frequency bands (delta, theta, alpha, and gamma), DAR, DTABR, and measures obtained during the EO condition, will be considered secondary or exploratory outcomes. For exploratory analyses involving multiple comparisons across frequency bands and conditions, false discovery rate correction using the Benjamini-Hochberg procedure will be applied to control for Type-I error inflation. All statistical analyses will be performed using SPSS Version 23.0 (IBM Corp., Armonk, NY, USA).

9.2. Statistical analysis plan for qEEG variables

Primary analysis will be beta-band frontal asymmetry. The primary analysis will evaluate group differences in beta-band frontal asymmetry measured during the EC resting-state condition. Depending on data normality, either an Independent Samples t test or the Mann-Whitney U test will be applied to compare the CPSP and non-CPSP groups.

Secondary and exploratory qEEG outcomes, absolute and relative power values for each frequency band (delta, theta, alpha, beta, gamma) will be computed from the Fp1 and Fp2 channels during both EO and EC conditions. Asymmetry features for the remaining frequency bands, as well as DAR and DTABR values, will be calculated accordingly. Between-group comparisons will be conducted using Independent Samples t tests or Mann-Whitney U tests, depending on the normality of data distribution. For analyses involving multiple frequency bands or conditions, false discovery rate correction will be applied as specified.

Linear Mixed Model (LMM) analysis, in addition to the primary and exploratory pairwise comparisons, LMM will be employed to account for the repeated-measures structure of qEEG data across resting-state conditions (EO vs EC) and frequency bands. In this model, groups (CPSP vs non-CPSP) will be treated as a between-subject factor, while condition and frequency band will be treated as within-subject factors. This modeling framework will allow evaluation of: (1) Overall group effects; (2) Group-by-condition interactions; and (3) Group-by-frequency interactions within a unified statistical model, thereby improving statistical efficiency and reducing inflation of Type-I error associated with multiple pairwise testing.

9.3. Additional and exploratory analyses

Additional analyses will be conducted to examine relationships between clinical characteristics such as medication use, NPSI scores, and PHQ-9 scores and qEEG indices. Because CNS-active medications (e.g., antidepressants, anticonvulsants including gabapentinoids, benzodiazepines, and other sedative-hypnotics) may influence EEG spectral characteristics, medication use will be considered a potential confounder. For the primary outcome, both unadjusted and medication-adjusted analyses will be presented. Medication variables will be incorporated into Analysis of Covariance or LMM models as prespecified binary indicators (e.g., CNS-active medication use: yes/no, or medication class-specific indicators), using a parsimonious adjustment strategy to minimize model overfitting given the relatively small sample size. If sample size permits, exploratory subgroup or sensitivity analyses will be conducted according to the presence or absence of CNS-active medication use. In addition, when recent medication changes (e.g., within 2 weeks prior to EEG acquisition) are identified, supplementary sensitivity analyses excluding these participants may be performed to assess the robustness of the findings. Furthermore, exploratory analyses may be performed for additional neurophysiological metrics including phase-amplitude coupling, weighted phase lag index, and the modified Brain Symmetry Index. These analyses will utilize the already acquired qEEG raw data and will not require additional EEG measurements. The decision to conduct such analyses will be based on data characteristics, interpretability, and expert consultation. Findings from exploratory analyses will be interpreted cautiously and used to complement the main results.

Discussion

CPSP is a significant complication that can severely impair both the physical and psychological quality of life of stroke survivors [1]. Previous studies have shown that patients with CPSP frequently experience psychological disturbances such as anxiety, depression, and sleep disorders, which may even lead to self-harm or suicidal behaviors in severe cases [1,14,16]. Following a stroke, CPSP is also associated with increased functional dependence and cognitive decline, thereby interfering with patients’ ability to perform independent activities of daily living, and this substantially contributes to socioeconomic burdens.

The current diagnosis of CPSP relies heavily on patient history, physical examination, neuroimaging findings, and the exclusion of other pain etiologies. No standardized diagnostic tools or criteria have been firmly established, and because CPSP often lacks overt neurological signs such as hemiparesis, it is frequently underestimated or diagnosed late. Therefore, establishing clear diagnostic criteria for CPSP is essential not only for accurate diagnosis but also for the objective evaluation of treatment outcomes [1].

Against this backdrop, qEEG has emerged as a promising methodology for evaluating CPSP. EEG offers millisecond-level temporal resolution which enables the detection of rapid changes in brain function, and is both noninvasive and cost-effective [6]. Previous studies have demonstrated the utility of qEEG in predicting stroke prognosis [14], evaluating post-stroke depression [16], and exploring neurophysiological features in chronic neuropathic pain patients [8]. In particular, indices such as the frontal asymmetry feature, DAR, and DTABR have been reported to reflect neural dysfunction, altered cortical excitability, and abnormalities in functional connectivity [6,8]. Among these indices, beta-band frontal asymmetry has been specifically associated with neuropathic pain severity and cortical hyperexcitability in prior studies [6,8]. Beta-band activity is thought to reflect alterations in sensorimotor integration and top-down modulation mechanisms, both of which are implicated in chronic neuropathic pain states. In light of its reported relationship with pain intensity and its use as the basis for sample size estimation in the present protocol for a study, beta-band frontal asymmetry under the EC condition was prespecified as the primary qEEG outcome in this protocol [6,8]. It has been reported that these findings provide an important foundation for investigating the potential of qEEG as an objective assessment tool for CPSP [8].

This study protocol possesses several methodological strengths. Firstly, by restricting enrollment to patients at least 3 months post-stroke, the unstable neurophysiological fluctuations, characteristic of the acute phase, were minimized, thereby improving the reliability of EEG measurements. Secondly, by focusing on frontal electrodes (Fp1 and Fp2), clinical feasibility was ensured whilst capturing meaningful EEG alterations related to pain processing. It has been reported that EEG signals recorded from the Fp1 and Fp2 electrodes reflect cortical surface activity and can provide valuable information regarding pain processing [8]. Although CPSP is primarily associated with dysfunction of the thalamocortical sensory pathway, pain perception is ultimately mediated by distributed cortical networks rather than isolated subcortical structures. The prefrontal cortex plays a pivotal role in both the perceptual and modulatory dimensions of pain, integrating sensory-discriminative and affective-emotional components. Previous studies have demonstrated that frontal EEG activity reflects alterations in pain-related processing including hemispheric asymmetry associated with lateralized emotional and sensory aspects of pain [8]. Notably, increased gamma-band activity in the prefrontal cortex has been linked to the subjective experience and chronicity of pain, suggesting that frontal oscillatory dynamics may serve as accessible markers of pain-related network dysfunction. In addition, thalamocortical dysrhythmia models propose that subcortical dysfunction can induce widespread cortical oscillatory alterations, which may be detectable at the scalp level. Therefore, even though frontal EEG does not directly measure deep thalamic activity, oscillatory indices derived from Fp1 and Fp2 may indirectly reflect thalamocortical network disturbances implicated in CPSP [8]. From a translational perspective, limiting EEG acquisition to frontal channels enhances clinical practicality. Multichannel EEG systems require considerable setup time and technical expertise, potentially limiting routine application. In contrast, frontal-channel-based algorithms may facilitate real-world implementation while retaining meaningful neurophysiological information. This balance between mechanistic relevance and clinical feasibility represents a key design consideration of the present protocol. Thirdly, by measuring both resting-state EEG (EO and EC) and cold-stimulation EEG, an evaluation of cold allodynia (a hallmark clinical symptom of CPSP) was incorporated. Fourthly, the use of well-validated qEEG indices including the asymmetry feature, DAR, and DTABR, reinforces the methodological rigor of the study by drawing on variables repeatedly demonstrated to be physiologically relevant in stroke and pain research.

The anticipated outcome of this research is to explore whether qEEG indices can serve as objective neurophysio-logical markers capable of discriminating between CPSP and non-CPSP patients. Such markers may facilitate early diagnosis, improve patient classification accuracy, and ultimately contribute to the development of personalized treatment strategies and reliable monitoring of therapeutic responses.

However, this protocol for a study also has limitations. As a single-center study with a relatively small sample size, the generalizability of the findings may be limited. In addition, the use of a 2-channel frontal EEG system restricts spatial resolution and may not capture the full extent of widespread neural network alterations. CPSP itself is influenced by multiple factors including lesion location, emotional and cognitive status, and medication use placing natural constraints on complete variable control. Furthermore, although the NPSI cut-off of ≥ 20 was adopted based on prior literature suggesting high sensitivity for CPSP identification [7], the relatively low specificity of this threshold may have introduced potential misclassification bias. As the NPSI is primarily a symptom severity scale rather than a standalone diagnostic instrument, future studies incorporating structured diagnostic criteria, such as the International Association for the Study of Pain or International Classification of Diseases-11 definitions of neuropathic pain, would strengthen diagnostic precision.

Despite these limitations, this study protocol represents an important study to investigate whether qEEG can distinguish CPSP from non-CPSP patients. By examining frontal EEG asymmetry, DAR, and DTABR under both resting-state and cold-stimulation conditions, the study may demonstrate the potential role of qEEG as an integrated neurophysiological tool for the diagnosis and evaluation of CPSP.

Article information

Author Contributions

Conceptualization: CHK. Methodology: CHK. Software: CHK. Validation: CHK. Formal analysis: CHK. Investigation: YS, HC, SJ and SK. Resources: CHK. Data curation: CHK. Writing - original draft preparation: YS. Writing - review and editing: CHK. Visualization: JK. Supervision: SL. Project administration: CHK. Funding acquisition: CHK. All authors have read and agreed to the published version of the manuscript.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Author Use of AI Tools Statement

The authors used ChatGPT (OpenAI) to improve the language clarity of the manuscript. All content was subsequently reviewed and edited by the authors, who take full responsibility for the final text.

Funding

This research was supported by a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute, funded by the Ministry of Health and Welfare, Republic of Korea (grant no.: RS-2025-02218421).

Ethics Statements

This protocol was reviewed and approved by the institutional review board of Wonkwang University Korean Medicine Hospital in Gwangju (WKIRB 2025/14-2, 17 September 2025).

Figure 1

Cerowave EEG recording device (Cerowave; Model: NGD-01)

(1) Velcro band: Adjustable Velcro strap that can be fitted according to the patient’s head size.

(2) Left measurement electrode: Channel 1 electrode attached to the left forehead.

(3) Right measurement electrode: Channel 2 electrode attached to the right forehead.

(4) Power button: Used to turn the device ON/OFF; pressing for more than 3 seconds activates the power, indicated by a blue LED light.

(5) Charging port: Micro 5-pin charging interface.

(6) Clamp: Mechanism that opens the electrodes by pressing, allowing proper placement for wearing.

(7) Reference/ground electrode: Reference and ground electrode attached to the right earlobe.

(8) Optical sensor: PPG sensor for pulse wave measurement.

EEG = electroencephalography; LED = light-emitting diode; PPG = photoplethysmography

Figure 2

EEG recording timeline

Prior to the initiation of EEG recording, participants underwent approximately 3 minutes of passive bilateral eye movements using an EMDR device for stabilization. Resting-state EEG was recorded for 5 minutes with EO, followed by an additional 5-minute recording with EC. Subsequently, cold-stimulation EEG was recorded for 2 minutes whilst applying a cold stimulus.

EC = eyes closed; EEG = electroencephalography; EMDR = eye movement desensitization and reprocessing; EO = eyes open

Figure 3

Study flow diagram

Arrows indicate the chronological flow of the study procedures.

References

• [1] Tamasauskas A, Silva-Passadouro B, Fallon N, Frank B, Laurinaviciute S, Keller S, et al. Management of central poststroke pain: systematic review and meta-analysis. J Pain 2025;26:104666. ArticlePubMed
• [2] Kim TI, Chung JW, Choi JW, Kim EJ, Lee JW, Kim YS, et al. A systematic review of treatment for chronic pain after stroke. J Int Korean Med 2016;37(6):929−39. [in Korean].ArticlePDF
• [3] Gwak JY, Cho SY, Shin AS, Lee IW, Kim NH, Kim HM, et al. Efficacy of bee-venom acupuncture on central post stroke pain: single-blind randomized controlled trial. J Korean Acupunct Moxibustion Soc 2009;26(6):205−14. [in Korean]. https://koreantk.com/ktkp2014/thesis/thesis-view.view?ctrlNo=HHCCB2_2009_v26n6_205.
• [4] GBD 2019 Stroke Collaborators. Global, regional, and national burden of stroke and its risk factors, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Neurol 2021;20(10):795−820.PubMedPMC
• [5] Choi HR, Aktas A, Bottros MM. Pharmacotherapy to manage central post-stroke pain. CNS Drugs 2021;35(2):151−60.ArticlePubMedPDF
• [6] Segning CM, da Silva RA, Ngomo S. An innovative EEG-based pain identification and quantification: a pilot study. Sensors (Basel) 2024;24(12):3873. ArticlePubMedPMC
• [7] Barbosa LM, da Silva VA, de Lima Rodrigues AL, Fernandes DTRM, de Oliveira RAA, Galhardoni R, et al. Dissecting central post-stroke pain: a controlled symptom-psychophysical characterization. Brain Commun 2022;4(3):fcac090. ArticlePubMedPMCPDF
• [8] Ryu S, Gwon D, Park C, Ha Y, Ahn M. Resting-state frontal electroencephalography (EEG) biomarkers for detecting the severity of chronic neuropathic pain. Sci Rep 2024;14:20188. ArticlePubMedPMCPDF
• [9] Lee H, Yang HL, Ryu HG, Jung CW, Cho YJ, Yoon SB, et al. Real-time machine learning model to predict in-hospital cardiac arrest using heart rate variability in ICU. NPJ Digit Med 2023;6:215. ArticlePubMedPMCPDF
• [10] Pagani M, Di Lorenzo G, Verardo AR, Nicolais G, Monaco L, Lauretti G, et al. Neurobiological correlates of EMDR monitoring - an EEG study. PLoS One 2012;7(9):e45753. ArticlePubMedPMC
• [11] Yun DJ, Oh J, Kim BJ, Lim JG, Bae JS, Jeong D, et al. Development of Korean neuropathic pain questionnaire for neuropathic pain screening and grading: a pilot study. J Korean Neurol Assoc 2012;30(1):15−25. [in Korean]. https://www.jkna.org/journal/view.php?number=5950&utm_source=chatgpt.com.
• [12] Kroenke K, Spitzer RL, Williams JBW. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med 2001;16(9):606−13.ArticlePubMedPMC
• [13] Finnigan S, van Putten MJAM. EEG in ischaemic stroke: quantitative EEG can uniquely inform (sub-)acute prognoses and clinical management. Clin Neurophysiol 2013;124(1):10−9.ArticlePubMed
• [14] Bentes C, Peralta AR, Viana P, Martins H, Morgado C, Casimiro C, et al. Quantitative EEG and functional outcome following acute ischemic stroke. Clin Neurophysiol 2018;129(8):1680−7.ArticlePubMed
• [15] Trotti A, Colevas AD, Setser A, Rusch V, Jaques D, Budach V, et al. CTCAE v3.0: development of a comprehensive grading system for the adverse effects of cancer treatment. Semin Radiat Oncol 2003;13(3):176−81.ArticlePubMed
• [16] Zheng YP, Wang FX, Zhao DQ, Wang YQ, Zhao ZW, Wang ZW, et al. Predictive power of abnormal electroencephalogram for post-cerebral infarction depression. Neural Regen Res 2018;13(2):304−8.ArticlePubMedPMC

Figure & Data

References

Citations

Citations to this article as recorded by