Application Mechanisms and Clinical Prospects of Brain–Computer Interface Technology in Radiation-Induced Brain Injury

Abstract

Radiation-induced brain injury (RIBI) is a common complication of radiotherapy that leads to neurological symptoms, significantly impairing patients’ daily functioning and long-term rehabilitation. Consequently, the development of effective therapeutic strategies has received considerable attention. As an emerging approach in the management of neurological disorders, brain–computer interface (BCI) technology shows substantial potential for both the assessment and treatment of RIBI. This review synthesizes evidence retrieved from electronic databases and examines RIBI from the perspectives of its mechanisms and clinical manifestations. Current findings suggest that BCI technology holds promise for several applications in RIBI, including early diagnosis, mitigation of neuroinflammation, alleviation of associated symptoms, and prediction and management of complications. The implementation of BCI is likely to play a significant role in early assessment and treatment processes for RIBI. Furthermore, with ongoing technological advancements, the development of next-generation BCI is expected to enable more targeted treatment that address additional pathological mechanisms of RIBI, thereby progressively improving the quality of life for affected patients.

Graphical Abstract

Introduction

Radiotherapy is one of the most widely used treatments for head and neck tumours and other malignancies (Owonikoko et al. 2014). Among its complications, RIBI is relatively common. It is a chronic disorder characterized by delayed and progressive neurological dysfunction (Gan et al. 2023). According to relevant studies, the incidence of RIBI after stereotactic radiosurgery for meningiomas ranges from 28% to 50%(Milano et al. 2018). The incidence of nasopharyngeal carcinoma after radiotherapy is approximately 3% (Na et al. 2014). In contrast, the incidence of brain metastases after radiotherapy is approximately 3 (Minniti et al. 2016), Moreover, the symptoms produced after treatment seriously affect patients’ quality of life, including neurocognitive impairment, motor dysfunction, brain edema, and radiation necrosis. Notably, cognitive impairment occurring more than 6 months after radiotherapy affects 50%–90% of patients (Makale et al. 2017), primarily manifesting as deficits in learning and memory.

Currently, research primarily focuses on symptomatic treatments; however, treatments targeting underlying pathological mechanisms remain limited, and significant challenges persist in both the evaluation and management of these conditions. Therefore, BCI technology may represent a promising approach to address gaps in the assessment and treatment of RIBI. As an emerging research technology, BCI has been widely used in the treatment of neurological diseases and to alleviate related symptoms. Some novel BCI technologies not only treat clinical symptoms but also alleviate symptoms and prevent complications by targeting underlying mechanisms. Therefore, based on an extensive literature review, we concluded that BCI has significant potential in RIBI, both for assessing and alleviating clinical symptoms and for treating related pathological mechanisms.

Literature Search Criteria

Starting from the PubMed and Web of Science databases, a comprehensive and systematic literature search was conducted from the establishment of these databases to April 2026. The search strategy integrated both keywords and Medical Subject Headings (MeSH terms), with the specific search keywords and MeSH terms including: radiation-induced brain injury, radiation-induced cognitive impairment, radiation necrosis, blood-brain barrier, neurogenesis, pharmacological intervention, neuroprotection, brain-computer interface, and radiotherapy.

During the search process, only full-text, peer-reviewed articles in English were included. Reviews, meta-analyses, original clinical studies, preclinical experimental studies, and relevant guidelines were considered eligible for inclusion. In contrast, case reports, conference abstracts, editorials, and non-English articles were excluded. Additionally, the reference lists of the included studies were manually screened to identify additional relevant publications that might not have been captured by the initial database search, ensuring the comprehensiveness and non-omission of the literature search and providing reliable evidence support for subsequent research.

Molecular and Cellular Mechanisms of RIBI

Radioactive brain injury can be divided into acute, early delayed, and late delayed injury phases (Tofilon and Fike 2000). Among these, the late delayed injury phase is the most critical. Different time phases exhibit distinct mechanistic changes and clinical manifestations; therefore, we will elaborate on these three phases separately, including their pathological mechanisms and clinical presentations, so as to provide an important entry point for the evaluation and treatment of BCI in the subsequent context. (Fig. 1).

Fig. 1.

Temporal phases and molecular mechanisms of radiation-induced brain injury. The pathophysiology of RIBI progresses through acute (< 2 weeks), early-delayed (1–6 months), and late-delayed (> 6 months) phases, characterized by distinct cellular and molecular events including BBB disruption, oligodendroglial apoptosis, demyelination, vascular injury, and impaired neurogenesis

Acute Phase (< 2 weeks)

The acute phase of RIBI, occurring within hours to weeks after irradiation, is characterized by rapid and profound disruption of the blood-brain barrier (BBB) and early loss of oligodendroglial progenitor cells (OPCs). Ionizing radiation triggers acid sphingomyelinase-dependent endothelial cell death within hours, leading to disorganization of tight junction proteins, particularly the downregulation of zonula occludens-1, and a marked increase in BBB permeability, which clinically manifests as vasogenic edema, increased intracranial pressure, headache, drowsiness, and exacerbation of pre-existing neurological deficits (Greene-Schloesser et al. 2012; Turnquist et al. 2020; Sharma and Czarnota 2019).

Concurrently, radiation induces an immediate but transient burst of intracellular reactive oxygen species (ROS) within minutes, which, despite normalizing within 30 min, initiates a cascade of mitochondrial dysfunction and sterile inflammation via the cGAS-STING pathway, further amplifying endothelial injury (Wang et al. 2025; Shimura et al. 2025). OPCs, the most radiosensitive glial cell population in this phase, undergo p53-mediated cell death within 8 h, while mature oligodendrocytes remain relatively preserved, setting the stage for subsequent chronic demyelination and white matter injury (Turnquist et al. 2020; Atkinson et al. 2005).

Early-Delayed Phase (1–6 months)

This stage involves reversible demyelination, somnolence syndrome, and cognitive impairment. Clinically, patients receiving radiation therapy may present with somnolence, nausea, anorexia, as well as deficits in short-term memory and attention. Somnolence syndrome usually improves within six weeks after the completion of radiation therapy (Powell et al. 2011). Molecularly, this phase features a second wave of BBB breakdown, spatially and temporally correlated with peak TNF production by activated microglia and astrocytes—vasogenic oedema leads to tissue hypoxia and ROS release (Liu et al. 2022). Neural precursor cells preferentially differentiate into glial cells rather than neurons, establishing a detrimental shift that persists (Monje et al. 2002a).

Late-Delayed Phase (> 6 months)

The late-delayed phase represents irreversible, progressive injury characterized by focal radio-necrosis, diffuse leukoencephalopathy, and cognitive decline. This phase manifests clinically as progressive deficits in memory, attention, executive function, processing speed, and, in severe cases, ataxia, urinary incontinence, and dementia. Focal necrosis presents with seizures, intracranial hypertension, and focal neurological deficits that can mimic tumour recurrence (Vellayappan et al. 2018; Crossen et al. 1994; Smith-Salzberg et al. 2025).

White matter is the predilection site of late-delayed RIBI, with primate transcriptomics showing the most prominent radiation-induced changes in deep white matter, not the hippocampus or cortex. The cascade begins with early oligodendrocyte precursor cell death, progressing to mature oligodendrocyte loss and demyelination months later, driven by impaired neural stem cell compensation (Lee and Tang 2022; Sokolov and Neumann 2013). Lipid metabolism disorders and progressive white matter changes (detected via DTI) further contribute to injury (Liu et al. 2025a). As illustrated, these changes—including reduced expression of glutamatergic receptors (SYP, GRIN2A, GRIA4)—lead to deep white matter-specific glutamatergic dysfunction and cognitive impairment independent of hippocampal injury (Andrews et al. 2019; Balentova and Adamkov 2015).

Late BBB breakdown precedes delayed radionecrosis and involves distinct mechanisms from acute disruption (Allen and Limoli 2022). VEGF-mediated vascular dysfunction is centrally driven by two key cascades: astrocytic hypoxia induced by vasogenic edema upregulates VEGF expression via HIF-1α signaling, while sustained neuronal unrepaired DNA damage triggers PARP hyperactivation, further promoting VEGF-A release and late barrier dysfunction (Wu et al. 2025). Additionally, astrocytic connexin-43 (Cx43) hemichannels mediate calcium-dependent VAMP3 vesicle trafficking and VEGF-A release at perivascular endfeet, directly inducing barrier leakage (Schumacher et al. 2024).

Hippocampal neurogenesis impairment is a key driver of radiation-induced memory deficits. First, radiation causes dose-dependent ablation of neural precursor cells, inducing cell death and leading to an immediate, long-lasting reduction in proliferating stem cells and immature neurons, while glial production remains unaffected (Cherng et al. 2026).Second, epigenetic silencing of BDNF via reduced HDAC1-dependent histone acetylation at its promoters suppresses BDNF transcription, impairing dentate gyrus neurogenesis and contributing to cognitive dysfunction, an effect reversible by HDAC inhibition (Ji et al. 2014). Third, neuroinflammation-mediated suppression driven by activated microglia and reactive astrocytes, most prominent in juvenile hippocampus, disrupts the neurogenic niche and inhibits neuronal production (Monje et al. 2002b; Raber et al. 2004). Finally, radiation triggers a shift in neural differentiation, with precursor cells preferentially adopting a glial fate, further reducing new neuron formation through both early-and late-delayed phases (Tofilon and Fike 2000).

Based on the aforementioned pathological mechanisms and clinical manifestations of RIBI, we propose that BCI-based assessment is suitable for early detection. Supported by relevant literature, acute and early delayed RIBI may be reversible injuries. Therefore, BCI-based therapy is more appropriately applied to irreversible late delayed brain injury induced by RIBI, aiming to alleviate related symptoms.

Current Limitations of Treatment Strategies

Although the current clinical treatment technologies for RIBI are diverse and well-established, they still have certain limitations and side effects. Therefore, we summarize the currently commonly used therapeutic strategies and their shortcomings, which provides partial theoretical support for the targeted application of BCI in the subsequent context.

Novel Therapeutic Strategies

Pharmacological interventions for radiation-induced cognitive impairment include memantine, which is recommended for consideration during whole-brain radiotherapy by NCCN guidelines but has only modest efficacy; in the RTOG 0614 trial, it failed to meet its primary endpoint of improved delayed recall (p = 0.059), with high attrition limiting assessment completion, though secondary analyses showed a reduced cognitive failure rate (53.8% vs. 64.9%), while optimal dosing remains challenging, particularly across different radiation modalities (Network 2025; Brown et al. 2013; Scampoli et al. 2022). Donepezil showed mixed results, failing to improve overall cognitive performance despite improvements in some individual memory tests, and psychostimulants such as methylphenidate and modafinil demonstrated benefits in small studies but were limited by low statistical power due to patient withdrawal, with NCCN guidelines noting limited overall evidence for these agents (Correa et al. 2016; Merchant et al. 2006). For symptomatic radiation necrosis, bevacizumab is the most evidence-based treatment, achieving 86% symptom improvement/stability and 93–94% radiologic response rates (Khan et al. 2021), though it is recommended only for corticosteroid-refractory cases and limited by cost, accessibility, and risks including thrombosis and pulmonary embolism; surgical interventions such as LITT or resection and hyperbaric oxygen therapy remain available but lack robust comparative data.

Critical Shortcomings of Current Strategies

The heterogeneity of RIBI complicates therapeutic development. While hippocampal dysfunction and loss of neural progenitor cells are recognized as key mechanisms, other cellular and molecular mechanisms appear less clear and remain to be elucidated as possible targets for treatment (Balentova and Adamkov 2015). The intricate crosstalk among direct cellular toxicity, inflammation, oxidative stress, mitochondrial dysfunction, and microglial activation drives a progressive downstream cascade that is difficult to interrupt with single-agent pharmacotherapy (Shamsesfandabadi et al. 2024). Compounding these issues, no validated biomarkers exist for early risk stratification or longitudinal monitoring of RIBI progression and treatment response (Sultana et al. 2020). Meanwhile, advanced imaging modalities including MR perfusion, spectroscopy, and amino acid PET suffer from suboptimal specificity and are further restricted by lesion size constraints, limited availability, and interference from treatment-related changes (Ali et al. 2019).

In contrast, BCI strategies bypass these limitations by directly modulating neural circuit function, enhancing signal transmission, and restoring cognitive performance without interfering with tumor control or relying on ambiguous biomarkers or heterogeneous molecular pathways. By targeting functional neural output rather than upstream pathological cascades, BCI offer a robust, tumor-safe, and mechanism-agnostic approach to alleviate RIBI-related cognitive deficits. Additionally, there are emerging targeted therapeutic approaches for inflammation alleviation, which also exert certain effects on the prevention and monitoring of other clinical manifestations and complications.

Basic Concept of BCI

A BCI is a communication system that establishes a direct link between neural activity and external devices, functioning independently of peripheral nerves and muscle output pathways (Vidal 1973). The advancement of BCI technology lies in its ability to bypass traditional neuromuscular pathways and provide a new method of communication for patients with motor dysfunction (Lebedev and Nicolelis 2006).

Since the first recording of brain electrical activity in 1924, BCI have undergone remarkable advances (Fig. 2). The technology has evolved from non-invasive BCI to invasive and semi-invasive types, and from open-loop operational principles to the gradually refined closed-loop systems used today. BCI continue to grow, constantly pushing the boundaries of neuroscience and rehabilitation.

Fig. 2.

Timeline of brain–computer interface development: landmark events from the first description in 1924 through clinical translation to today’s broad applications

Working Principle of BCI

The working principle of a BCI involves four key components: signal acquisition, signal processing and feature extraction, signal decoding, and feedback and control (Schalk et al. 2004) (Fig. 3).

Fig. 3.

Brain–computer interfaces are categorized into three types based on electrode location. The key steps of their working principle are as follows

Signal Acquisition

This stage employs various BCI electrode technologies to capture neural electrical activity for subsequent extraction and decoding.

Signal Processing and Feature Extraction

Raw neural signals are preprocessed to remove artefacts and extract task-relevant features. Feature extraction is traditionally classified by electroencephalographic signatures: Time-domain—pioneered by Farwell et al., who used event-related potential (ERP) peak latency to drive a P300 speller, is now the canonical time-domain paradigm (Farwell and Donchin 1988). Frequency-domain—band-power changes are exploited in motor-imagery and steady-state visual evoked potential (SSVEP) BCI to decode user intent. Spatial-domain—distribution-specific patterns predominantly applied in motor-imagery paradigms.

Signal Decoding

Signal decoding forms the core of BCI signal processing. Commonly employed algorithms include canonical correlation analysis (CCA), common spatial patterns (CSP), and linear discriminant analysis (LDA).

Control and Feedback

In Open-loop BCI systems, decoded neural commands are directly transmitted to external devices. Robotic arms remain the most common actuators. The integration of Artificial Intelligence (AI) and Robotic arms enables tetraplegic users to achieve refined, sustained control of robotic arms over prolonged periods (Natraj et al. 2025). BCI have also been integrated with computer peripherals, such as cursors and spellers, to restore communication. Willett et al. successfully decoded handwriting intentions via BCI. They converted them into text with 99% accuracy using a cursor-based interface (Willett et al. 2021).

Functional electrical stimulation (FES) is emerging as a promising therapeutic modality. Lorsch et al. introduced the brain–spine interface (BSI), which enables spinal-cord-injured patients to regain over-ground walking and promotes neural plasticity (Lorach et al. 2023). Feedback mechanisms close the BCI loop, enabling adaptive self-adjustment based on decoded signals. This is an essential direction for future BCI development.

The most widely used classification of BCI is based on the method used to acquire signals. According to this criterion, BCI are generally divided into three categories: invasive, semi-invasive, and non-invasive systems, each differing in the degree of physical penetration into neural tissue and the resulting signal fidelity, safety, and application scope.

Classification of BCI

Invasive Brain-Computer Interfaces (iBCI)

iBCI use surgically implanted microelectrode arrays to record single-or multi-unit neuronal activity directly. A landmark advance in intracortical recording hardware was the development of the tungsten micro-wire electrode by Hubel et al. Its high mechanical strength allowed penetration of the dura mater and laid the groundwork for subsequent rigid-shaft electrodes (Hubel and Wiesel 1962). Building on this precedent, the Utah and Michigan arrays emerged as the two dominant silicon-based technologies and remain pivotal in current iBCI.

The Utah array’s principal advantage is its ability to simultaneously sample multiple neurons, which dramatically increases the channel count. In 1988, Drake et al. reported the first in vivo application of the Michigan probe, marking its debut as an implantable neural interface (Drake et al. 1988). Contemporary iBCI now incorporate wireless telemetry and employ ultra-flexible, minimally injurious electrodes to mitigate chronic inflammation and signal loss (Simeral et al. 2021; Gu et al. 2021).

Semi-Invasive BCI

Semi-invasive BCI primarily include electrocorticography (ECoG) and intracranial EEG (iEEG), with ECoG dominating current research. Penfield and Jasper described the first ECOG for localizing epileptic foci (Penfield and Jasper 1954). Leuthardt et al. introduced ECoG as a BCI modality termed BCI-ECoG, in which electrodes are placed between the scalp and the cortical surface, abutting but not penetrating the dura (Leuthardt et al. 2004). This configuration has a lower surgical risk than fully invasive approaches and provides higher signal fidelity than non-invasive recordings.

Building on this foundation, modern flexible ECoG technologies now employ advanced polymers and micro-machined architectures that enhance signal-to-noise ratio and markedly reduce peri-implant inflammation (Hu et al. 2024; van Oosterhout et al. 2025).

Non-Invasive BCI

Non-invasive BCI acquire neural signals through scalp-surface electrodes or optical techniques such as functional near-infrared spectroscopy (fNIRS). These approaches are painless and easy to apply, but they inherently suffer from lower signal-to-noise ratios (Wolpaw et al. 2002).

Scalp electroencephalography (EEG) electrodes can be categorized by morphology into three types. Berger et al. performed the first human EEG using silver–silver-chloride disc electrodes with conductive paste, which represented the original “wet” electrode (Berger 1929). To overcome the inconvenience of gel application while preserving signal fidelity, Gramatica et al. developed dry electrodes (Gramatica et al. 2006). Because dry electrodes exhibit higher impedance and wet electrodes are cumbersome, Wang et al. introduced semi-dry electrodes in 2016 (Wang et al. 2016).

Recently, the quest for simpler yet high-precision electrodes has driven research into flexible non-invasive interfaces. Ultra-flexible tattoo electrodes, for example, are stretchable and easy to apply while preserving signal fidelity (Wei et al. 2023). Furthermore, it represents the current frontier of non-invasive BCI hardware.

Current Applications of BCIs

Treatment of Motor Dysfunction

BCI are increasingly used to address limb motor dysfunction arising from conditions such as spinal cord injury, post-stroke paralysis, and amyotrophic lateral sclerosis (ALS). Their applications primarily focus on replacing motor function and enhancing neural plasticity.

Motor function-replacement BCI have advanced more rapidly. Following signal acquisition and decoding, these systems can directly control external devices to compensate for impaired limb function. For example, Handelman et al. used bilateral implanted BCI to enable paralyzed patients to feed themselves independently using dual robotic arms (Handelman et al. 2022). Similarly, Donati et al. observed partial recovery of neural function in patients with paraplegia through long-term gait training mediated by BCI (Donati et al. 2016). As a result, BCI can promote neural plasticity through repeated brain stimulation, facilitating the re-establishment of some neural connections and the partial restoration of motor function.

Restoration of Speech Function

Currently, the fastest-developing approach is to use BCI to directly decode speech signals, enabling patients to “speak” through thought—for example, as in Metzger et al.used ECoG to directly decode patients’ speech signals, allowing them to spell words (Metzger et al. 2022). Willett et al. implanted electrodes in the motor cortex to directly record neural activity and decode it, significantly improving patients’ vocabulary accuracy (Willett et al. 2023). By using BCI, patients can achieve substantial restoration of speech function, thereby enhancing communication and quality of life.

Integration with Other Technologies

The combination of BCI with other emerging technologies is enhancing treatment outcomes. Some studies have demonstrated that integrating BCI with functional electrical stimulation (FES) can improve patients’ limb function more effectively than either technology alone (Cantillo-Negrete et al. 2023). Putze et al. combined BCI with augmented reality (AR) technology, thereby increasing AR immersion and enhancing BCI rehabilitation outcomes (Putze et al. 2020). Such integrations are becoming an increasingly prominent focus in BCI research.

The Dual Role of BCI in RIBI

At present, the difficulty in evaluating RIBI lies in the inability to assess cognitive function accurately. Moreover, severe cognitive dysfunction can also affect the application of imaging techniques. In terms of treatment, cognitive dysfunction is poorly addressed, and the reduction of brain inflammation is also unsatisfactory. Therefore, we believe that BCI technology can address the limitations outlined above. To explore this potential, we conducted a more systematic analysis of this approach.

Potential of BCI in Assessing RIBI

Among the sequelae of radiation-induced brain injuries, long-term cognitive decline is a primary focus of ongoing research. Cognitive decline can significantly impair patients’ daily functioning and quality of life. However, current methods for assessing cognitive function remain neither convenient nor entirely accurate.

Non-invasive BCI offer a promising approach for the in-depth analysis of neural signals, enabling objective assessment of patients’ cognitive function and potentially supporting long-term monitoring as a prospective evaluation method (Fig. 4). The following section provides a detailed explanation of this approach.

Fig. 4.

The Application of Brain-Computer Interfaces in Assessing Radiation-Induced Brain Injury: (1) To assess cognitive impairment. (2) Integrate additional neuroimaging modalities. (3) To predict the complication of epilepsy and enable early intervention

Assessing Cognitive Function

RIBI can lead to cognitive decline in affected patients. Assessing cognitive function is also a major challenge in RIBI. Traditional evaluation scales or imaging techniques are used only when symptoms are relatively severe, and patients can sometimes influence these scales. Therefore, we have found that EEG, which monitors brain electrical activity, offers a potential means not only to assess overall brain function affected by radiation-induced brain damage but also to evaluate cognitive function, for example, as in Damji et al. investigated EEG microstates as potential biomarkers of changes in cognitive-related brain dynamics following breast cancer chemotherapy (Damji et al. 2025). Radiation-focused investigations have demonstrated that both EEG entropy and brain microstates are altered after exposure, collectively indicating that electroencephalography is a valuable tool for identifying and quantifying RIBI.

This kind of assessment, which focuses on the microstates of cognitive function, is most distinct from traditional imaging techniques in that it can detect changes in cognitive function before obvious symptoms appear, facilitating early treatment. It is also more accurate than the results obtained from scale assessment.

Multi-Technology Integration

BCI technology can be integrated with multiple imaging modalities, such as fMRI, PET, and fNIRS, to establish a multimodal assessment system. The integration of this technology can compensate for the limitations of traditional imaging techniques. For example, fNIRS can measure cerebral blood flow and metabolic changes, while combining EEG with fNIRS provides a more comprehensive assessment of brain function and structure, compensating for the limitations of single-modality systems (Lee et al. 2015). Similarly, integrating EEG with fMRI combines the high temporal resolution of EEG with the high spatial resolution of fMRI, overcoming the inherent constraints of using either modality alone (Lioi et al. 2020). The latest BCI technology, LiTA-HM, can monitor blood flow throughout the brain and detect Hypoxia in blood vessels (Liu et al. 2025b). This provides the possibility of evaluating the mechanism of RIBI at a fundamental level.

These multimodal approaches to RIBI enable clinicians to make more accurate assessments of cognitive and neurological symptoms and to generate robust data to support targeted treatment.

Real-Time Monitoring and Early Warning

Epilepsy is a serious complication of RIBI and has long been a research focus in early monitoring and prevention. However, treatment techniques for epilepsy caused by RIBI after its onset are relatively mature, but early monitoring and early warning remain limited. Therefore, to avoid sudden epileptic seizures that can affect the recovery of patients with RIBI, we have found that the current BCI can play a role.

Acharya et al. utilized automated EEG techniques to predict epileptic seizures several minutes to hours in advance by analyzing features such as increased EEG δ-band power and elevated high-frequency oscillations (Acharya et al. 2013). They also found that automated detection of high-frequency oscillations can aid in localizing epileptic foci. Similarly, Kashmiri et al. used a partially invasive Ekong-based BCI to analyze the Lyapunov exponent (L). They found that it could detect and localize epileptic seizures through multi-electrode analysis (Iasemidis et al. 1990).

The findings of these studies fully demonstrate that it is possible to predict epilepsy at an early stage. At the same time, they improve the prognosis for RIBI and reduce the impact of sudden complications.

Feasibility of BCI in Treating RIBI

By targeting the underlying pathophysiological mechanisms, BCI may help mitigate neurological deficits and improve associated clinical symptoms (Fig. 5).

Fig. 5.

Brain-Computer Interface for the Treatment of Brain Inflammatory Response and Dysfunction Caused by Radiation-Induced Brain Injury: (1) Use drug-coated devices or drug-loading carriers to alleviate intracranial inflammation. (2) To treat cognitive impairment—for example, by using BCI for monitoring while combining it with techniques such as tDCS. (3) To treat motor dysfunction that may result from radiation-induced brain injury

Applications of BCI in Modulating Inflammatory Responses

Radiotherapy can trigger inflammatory responses in the brain, leading to the release of pro-inflammatory factors, vascular damage, and early tissue injury. Controlling inflammation within the brain has therefore become a key aspect of treating RIBI. In recent years, studies have proposed using flexible deep BCI in combination with anti-inflammatory drugs and coated carriers.

For intracranial inflammation, the most commonly used treatment is medication. However, long-term drug use can easily lead to patient dependence and infection. Therefore, some studies have proposed delivering anti-inflammatory drugs via BCI technology to alleviate intracranial inflammation. This method can directly target the affected area, cause fewer side effects and achieve better therapeutic effects than traditional drug treatment. For example, medications such as dexamethasone can alleviate neuroinflammation (Lv et al. 2025). Qin et al. simultaneously recorded ECoG signals while using a flexible BCI electrode with antibacterial and anti-inflammatory properties. The electrode, loaded with anti-inflammatory drugs embedded in a bacterial cellulose hydrogel, allows for slow drug release through the BCI electrode and effectively suppresses bacterial growth and neuroinflammation (Qin et al. 2024). These findings provide preliminary evidence for the feasibility of using BCI to reduce neuroinflammation caused by RIBI.

Role of BCI in Cognitive Dysfunction

Radiotherapy can cause neuronal damage, impairing cognitive function by inhibiting the differentiation of mature neurons. The current treatment methods mainly fall into two categories: one is drug therapy, and the other is hyperbaric oxygen therapy. However, the prognosis for both treatment methods has not improved significantly. BCI fill this gap by providing a closed-loop treatment approach that integrates with other therapeutic techniques to enhance patients’ cognitive function.

Non-invasive BCI technologies have been shown to promote neuroplasticity by targeting specific brain regions. For example, as shown by Gomez-Pilar et al. found that neurofeedback therapy (NFT) and motor imagery-based BCI (MI-BCI) improved cognitive function in older adults (Gomez-Pilar et al. 2016), suggesting the potential to restore cognitive function in patients with RIBI.

Similarly, Tabikh et al. combined non-invasive BCI with iEEG and fast periodic visual stimulation (FPVS). When specific changes in brain signals related to cognitive functions are detected, transcranial direct current stimulation (tDCS) applied in this configuration will generate an electrical current, resulting in substantial improvements in the patient’s cognitive functions (Tabikh et al. 2025). These studies indicate that BCI can serve not only as a marker for brain monitoring but also as a tool to advance research into improving cognitive functions.

BCI for Restoring Limb Motor Function

MI-BCI are currently among the most widely used non-invasive BCI technologies for rehabilitating limb function. Based on neural closed-loop theory, MI-BCI establishes a central-peripheral-central rehabilitation system that enhances network-level brain signaling and facilitates recovery of limb motor function (Kruse et al. 2020). When combined with tDCS, this approach produced measurable improvements in upper-limb motor function. BCI-based technologies can collect brain signals and guide the development of individualized rehabilitation plans, offering a promising strategy for addressing motor deficits resulting from RIBI.

Prospects for Future Research

AI/ML-empowered digital twins: AI-and machine learning–driven digital twin frameworks will leverage deep learning algorithms to curate multimodal RIBI datasets and construct patient-specific digital brain models. These models will enable forecasting of disease trajectories, symptom progression, and treatment responses. In parallel, AI-driven signal optimization will facilitate closed-loop, adaptive neuromodulation protocols tailored to individual radiosensitivity profiles, enabling precision neurotherapeutic treatment methods.

Applications of Emerging BCIs: BCI technologies are rapidly advancing and have demonstrated remarkable efficacy in animal models. Notably, the development of organic BCI (OBCI) has enabled reconstruction of damaged neural circuits, leading to functional restoration. If translated into RIBI therapies, such technologies may yield significant therapeutic benefits. In the context of epilepsy prediction, emerging evidence suggests that integrating BCI with AI-based analytics can enhance predictive accuracy and enable earlier detection. Collectively, these advances provide valuable translational insights for the future clinical management of RIBI.

Early Prediction of RIBI: A major clinical challenge in RIBI remains the inability to predict disease onset at the individual patient level. Recent advances in BCI technology, such as LiTA-HM–based full-brain vascular monitoring, offer a potential solution. This platform could enable systematic characterization of cerebrovascular differences between patients who develop RIBI and those who do not. Identification of reproducible vascular biomarkers could provide critical predictive indicators for early RIBI risk stratification and preclinical treatment.

Discussion

RIBI remains a significant barrier to improve the long-term quality of life of cancer survivors. Its complex pathophysiology involves neuronal cell death, impaired neurogenesis, vascular injury, and persistent neuroinflammation, which together lead to severe cognitive and functional decline. Current treatments offer only partial relief, underscoring the need for innovative solutions. BCI technology represents a promising frontier, enabling real-time assessment, seizure prediction, cognitive rehabilitation, motor recovery, and even modulation of neuroinflammatory processes. While challenges remain, integrating BCI into clinical practice has the potential to transform the management of RIBI and significantly improve patient outcomes.

Although BCI technology offers significant potential for the treatment of RIBI, its limitations must not be overlooked. The irradiated brain exhibits reduced tolerance to external perturbations; consequently, RIBI requires BCI that are effectively atraumatic. However, current non-invasive and semi-invasive systems still compromise spatial resolution and signal-to-noise ratio. Moreover, early-stage RIBI is characterized by subtle or subclinical lesions, resulting in a scarcity of usable neural signals for conventional EEG and ECoG platforms. The field is therefore moving toward wireless, flexible, high-density interfaces that integrate a near-zero surgical footprint with single-unit-level recording fidelity.

Implanted electrodes elicit foreign-body responses that are likely exacerbated in the irradiated, immunocompromised brain. In patients undergoing long-term low-dose stereotactic radiotherapy for life prolongation, implanted hardware is further exposed to chronic inflammation, fibrosis, and microvascular compromise. Future clinical trials must therefore incorporate radiobiological safety endpoints alongside conventional efficacy metrics to ensure long-term signal stability, biocompatibility, and device integrity in this uniquely vulnerable population.

Although this article discusses the optimism surrounding BCI applications, there is currently no practical evidence of its use in RIBI. We discuss its potential application in RIBI, drawing on BCI applications in other diseases and basic medical research, and suggest that it may be useful in RIBI. However, specific investigations through experiments or clinical studies are still required. For example, during RIBI recovery, scar tissue may form in the brain, potentially affecting BCI signal transmission and the duration of BCI use.

Author Contributions

Conceptualization: Ruijun Xu, Xueqian Huang, Zhanlin Chen, HUSSEIN ALI MOHAMED, Xuyu He, Feiyang Lu, YSO, Ge Li, Kunyi Zhang; Data curation: Ruijun Xu, Xueqian Huang, Zhanlin Chen; Formal analysis: Ruijun Xu, Xueqian Huang, Zhanlin Chen, HUSSEIN ALI MOHAMED, Xuyu He, Feiyang Lu, YSO; Methodology: Xuyu He, Feiyang Lu, Ge Li, Kunyi Zhang; Project administration: Xuyu He, Ge Li, Kunyi Zhang; Visualization: Ruijun Xu, Xueqian Huang, Feiyang Lu; Writing – original draft: Ruijun Xu, Xueqian Huang, Xuyu He, Ge Li, Kunyi Zhang; Writing – review & editing: Ge Li, Kunyi Zhang.

Funding

This work was supported by the Natural Science Foundation of Guangdong Province (2024A1515011451), Li Xin Traditional Chinese Medicine Research and Innovation Fund (2023-CCA-TCM-033), and the Wohua Research Fund (BYPDF2411213), the National Natural Science Foundation of China (82271395), the Science and Technology Development Fund of Macau (0011/2025/RIA1), the Guangdong Basic and Applied Basic Research Foundation (2023A1515030073), the Guangzhou Science and Technology Plan Project (2025A04J4740), Excellence Initiative Project of the National Natural Science Foundation of China (KY012026190), and the Special Project of Dengfeng Program of Guangdong Provincial People’s Hospital (KY0120220133 and DFJHBF202111).

Data Availability

No datasets were generated or analysed during the current study.

Declarations

Competing Interests

The authors declare no competing interests.

Consent to Participate

Informed consent was obtained from all individual participants included in the study.