Functional mapping of the cerebral cortex in clinical practice. Modern technologies and development prospects

Full Text

Introduction

Despite centuries of research, the functional organization of the cerebral cortex remains a subject of intense scientific interest. Since the beginning of the 21^st century, over 1,000 scientific articles on this topic have been published annually, and the debate between localizationists and holists initiated by F. Goltz and D. Ferrier in 1881 continues to this day [1].

Recent decades have witnessed rapid advances in microelectronics, computing, and materials science, leading to the development of brain–computer interfaces, closed-loop neurostimulators, and neuroprostheses that communicate directly with the cerebral cortical surface. This has underscored the need for a highly precise understanding of the principles of brain function, functional localization, and their dynamic interrelationships.

Functional cerebral cortex mapping (FCCM) has become firmly integrated into routine clinical practice; however, no ideal method that could rightfully be termed the gold standard has yet been developed.

The study aim is to systematize knowledge of FCCM technologies currently used in clinical practice in patients with supratentorial brain tumors and drug-resistant epilepsy, as well as in neurorehabilitation programs. A critical assessment of their diagnostic yield is presented, and directions for future development are formulated.

We did not analyze such FCCM methods as fluorodeoxyglucose positron emission tomography (FDG-PET), magnetoencephalography (MEG), high-density electroencephalography (hdEEG), or near-infrared spectroscopy (NIRS), as their diagnostic utility is still under investigation and their application in clinical medicine remains limited. Intraoperative localization of the central sulcus by phase reversal of somatosensory evoked potentials was not included in this study, as it has long been standardized and has a narrow range of practical applications.

Invasive Mapping Methods

Observations first published in 1870 by G. Fritsch and E. Hitzig — namely, movements of the contralateral limbs of a living dog at the moment a galvanic current was passed through certain areas of the cerebral cortex — initiated the development and study of direct cortical stimulation (DCS) for functional mapping [2].

The first successful clinical application of DCS in neurosurgery was the localization and subsequent resection of a brain area in a patient with focal epileptic seizures involving the face [3]. From the earliest observations of the effects of electricity on brain tissue, it became clear that this could produce one of three reactions:

• excitation (activation), leading to positive responses (e.g., muscle contractions);
• inhibition, causing temporary functional disruption (e.g., speech arrest);
• hyperexcitation, manifesting as an epileptic seizure.

Whether performed intraoperatively or extraoperatively, direct DCS remains the gold standard for localizing functionally significant areas of the brain, including white matter tracts.

Intraoperative Electrical Cortical Stimulation

The practical implementation of DCS for FCCM was advanced by the surgeries of W. Penfield, a pioneer of functional neurosurgery. The method he developed for treating focal epilepsy, termed the Montreal Procedure, was based on electrical stimulation of the brain using a bipolar fork-shaped electrode with faradic current delivered at a frequency of 60 Hz, accompanied by simultaneous electrocorticography (ECoG) recording [4]. During surgery, patients remained awake, and local DCS was used to provoke a typical epileptic seizure as a biomarker of the epileptogenic zone to be resected. A byproduct of this approach was detailed mapping of motor and sensory functions of the cerebral cortex, subsequently visualized as the Penfield homunculus [5]. A crucial finding was the practical confirmation that functional localization in the cortex is relatively individualized, although concentrated in specific regions that are similar across individuals.

Currently, intraoperative electrical stimulation mapping of the cortex and cerebral white matter tracts is performed during the majority of neurosurgical procedures for supratentorial tumors and tailored resections of epileptogenic zones; however, debates regarding stimulation methodology remain relevant. A meta-analysis of 19 studies encompassing 922 patients showed that despite DCS-guided FCCM, focal postoperative neurological deficits persist in 19% of patients, of which 72% do not fully recover [6]. This relatively modest outcome may be explained by several factors but, in any case, raises the question of whether the method should be accepted as the gold standard against which other approaches should be compared.

The expansion of the electrical stimulation area around the electrode with increasing current intensity, along with the risk of DCS-induced epileptic seizures, explains the necessity of determining the threshold current intensity [7]. This process is relatively simple and reproducible for motor mapping; however, determining thresholds for localization of cognitive functions is challenging. It is generally accepted that non-motor mapping in awake patients should be performed at current intensities 0.5–1.0 mA below the motor threshold, although this parameter was derived empirically [8], despite significant variability in cortical excitability across different regions within the same patient [9].

During his surgeries, W. Penfield used ECoG not only to detect epileptiform discharges as a sign of an irritative area but also to detect the onset of epileptic seizures in order to differentiate them from the DCS effect on the mapped area. It is believed that afterdischarges on ECoG (epileptiform activity occurring immediately after stimulation) indicate that the current intensity threshold has been reached [10]. Clinical experience indicates that such afterdischarges do not always correlate with increasing current intensity, and sometimes clinical signs of an epileptic seizure may occur even before epileptiform activity appears on ECoG. This is likely due to the propagation of the afterdischarge wave and/or epileptiform activity from the stimulating probe away from the cortical recording electrode. To avoid this effect, W.O. Tatum and colleagues proposed using a ring-shaped ECoG electrode placed around the area being mapped; however, to date, such an electrode is not available in catalogs of consumables approved for intraoperative use [11].

In practice, two approaches to intraoperative mapping exist: positive mapping and negative mapping. In the first approach, localization of a functional area is mandatory to confirm the adequacy of the selected stimulus intensity, followed by mapping of the resection area; however, this takes significantly more time and necessitates a large craniotomy, which always increases the risk of surgical complications. In the second approach, the absence of a clinical effect from DCS is considered sufficient justification to resect the corresponding cortical areahowever, the absence of an effect may result from insufficient current intensity, a poorly chosen cognitive task, or a subclinical epileptic seizure, any of which can produce a false-negative result.

Extraoperative Electrical Brain Stimulation

The development of polymeric materials for flexible, tissue-neutral electrodes and digital electroencephalographs with large storage capacity has led to the emergence of multi-day invasive video-electroencephalography monitoring. The transition of the preparatory stage of surgical treatment for focal epilepsy from the operating room to the hospital room has changed the approaches to DCS for FCCM, as time constraints on procedures have been eliminated, the patient is always in a comfortable position, and can rest if necessary. This has enabled comprehensive FCCM of non-native language speakers, mapping of various cognitive functions, and monitoring of ictal state dynamics during DCS. Extraoperative DCS for FCCM can be used as an alternative to awake surgery, for example, in cases where awake surgery is contraindicated due to the patient’s cognitive deficit or psychological state.

Currently, two types of invasive electrodes are used:

• Subdural electrodes — in the form of flat strips or grids, placed through a craniotomy window under the dura mater (subdural ECoG);
• Depth electrodes — in the form of long, thin cylinders, inserted into brain tissue using a stereotactic frame or robot (stereo-EEG).

Both techniques were developed in the 1960s [12]. Unlike subdural ECoG, which allows assessment of only the convexital and basal cortical surfaces, stereo-EEG records bioelectrical activity with high spatial resolution throughout the entire volume of the telencephalon, including its mediobasal and other difficult-to-access regions, within the boundaries of the implanted electrodes [13]. In both cases, DCS through the implanted electrodes is used both for provoking typical epileptic seizures and for FCCM.

For over half a century, subdural ECoG and stereo-EEG have competed in the presurgical workup for surgical treatment of epilepsy. A significant advantage of subdural electrodes is the simplicity of FCCM and its methodological similarity to the intraoperative approach. However, a comparative analysis of the FCCM utility through depth electrodes has shown that its accuracy is not inferior to that of DCS via subdural strips/grids [14, 15].

The choice of DCS technique is important for clinical decision-making in the surgical treatment of drug-resistant epilepsy and brain tumors. The Table 1 summarizes the advantages and disadvantages of each approach to FCCM using DCS.

 

Table 1. Comparison of DCS application features for FCCM

Characteristic	Intraoperative FCCM with handheld probe	Subdural ECoG	Stereo-EEG
Surgical access	Wide craniotomy	Craniotomy with reoperation for electrode removal	Burr holes with diameter matching the implanted electrode
Brain coverage (superficial/deep)	Only accessible cortical surface within craniotomy limits	Convexital cortical surface, basal temporal lobe regions	Superficial and deep structures, including sulcal fundi, insula, cingulate gyrus, medial temporal structures; capability for bilateral investigation
Testing duration and conditions	Limited by operative time, depends on patient cooperation	Several days, in hospital room setting, repeat testing possible	Several days, in hospital room setting, repeat testing possible
Advantages	Single-stage procedure (diagnosis and resection in one operation), direct visual control of stimulation sites	Ability to fully cover large areas of the convexital cortical surface	Minimally invasive, mapping of deep structures, bilateral investigation, higher safety profile
Limitations	Time and spatial constraints, effects of anesthesia, dependence on patient cooperation	High invasiveness, inability to map deep structures, relatively high complication risk	Sparse coverage of the cortical surface
Typical clinical scenarios	MRI-positive epilepsy or tumors in eloquent areas with clear localization hypothesis	Epilepsy or tumors with suspected focus on the convexital brain surface	Epilepsy or tumors with suspected focus in deep structures, or when bilateral investigation is required

 

Despite its formal status as the gold standard, DCS protocols for FCCM remain poorly standardized, leading to heterogeneity in practice across different centers. Nevertheless, commonly accepted ranges for stimulation parameters exist: pulse width (0.2–1.0 ms), frequency (50 or 60 Hz), train duration (3–5 s), and current intensity (1–20 mA) [16]. Differences in stimulation parameters for ECoG and stereo-EEG are determined by the physical characteristics of the electrodes and the objectives of the study. A key parameter is charge density — the product of current intensity and pulse duration, divided by the surface area of the electrode contact with brain tissue. Charge density determines both stimulation efficacy and the risk of thermal tissue damage, making it a fundamental metric for balancing data acquisition with patient safety. For example, due to differences in electrode geometry and significant current shunting through cerebrospinal fluid (up to 87% for subdural electrodes), lower current intensities are required for stereo-EEG to achieve comparable charge density in brain tissue.

Functional Electrocorticography

Observable changes in the amplitude and frequency of bioelectrical activity recorded by ECoG from the cerebral cortical surface during a speech act were first demonstrated in a publication by G.A. Ojemann et al. [17]. N.E. Crone et al. showed that the biomarker of a speech functional area is an increase in power in the γ-band frequency range exceeding 80 Hz [18]. Since that time, intensive investigation has been underway into the potential of functional ECoG, based on recording changes in γ-frequencies in response to cognitive task performance, for mapping functional areas without the use of electrical stimulation. For simplicity, the term has been abbreviated to functional ECoG (fECoG). The potential ease of fECoG use in brain–computer interfaces lends particular relevance to developments based on this technique, since all systems that use invasive connections to the brain are based on quantitative ECoG analysis [19]. For clinical application, advantages include the complete absence of induced epileptic seizures during mapping — which is particularly important in patients with epilepsy — as well as the ability to spatially assess the changes in the entire speech act.

fECoG can be performed both during awake neurosurgical procedures on the cerebral cortex and in patients with implanted electrodes undergoing presurgical evaluation for drug-resistant epilepsy. It should be noted that for functional mapping based on evoked γ-activity on the corticogram in response to a cognitive task, subdural electrodes are predominantly used, due to the ability to achieve broader cortical coverage compared to depth electrodes. The methodology involves the presentation of stimulus tasks (typically images of objects and/or actions, analogous to mapping with electrical stimulation) and synchronous recording of bioelectrical activity epochs that include a 1000 ms pre-stimulus baseline and 2000 ms post-stimulus period. The recorded signal undergoes mathematical processing, and as responses are averaged, a heat map reveals the electrodes in which an increase in γ-band frequency power is observed during word production (Fig. 1).

 

Fig. 1. Graphical interface of the EloQ server application for fECoG.

A — time series window displaying native ECoG signals by channel; B — audio channel window; C — functional task display window; D — heat map.

 

Despite the common neurophysiological principle of procedure, algorithms for ECoG signal analysis vary. The most widely used protocol is SIGFRIED (SIGnal modeling For Real-time Identification and Event Detection), first introduced in 2009 [20]. Despite the age of this development and an important limitation — the necessity of recording resting-state ECoG before mapping begins — a large number of publications have been devoted to it. This is due to the implementation of this protocol in the CortiQ device (G.tec medical engineering GmbH), which has received clinical approval in many countries [21]. In Russia, the first clinical trials of a device for cortical mapping using fECoG without electrical stimulation have been completed. This device, named EloQ (an acronym for Eloquent Cortex), demonstrated an area under the ROC curve of 0.91 in extraoperative patients with implanted strip electrodes [22].

A meta-analysis by R. Arya et al. demonstrated relatively low sensitivity and specificity for fECoG — 61% and 79%, respectively [23]. However, the authors included both extraoperative and intraoperative studies conducted using various stimulus presentation paradigms in different languages and analyzing multiple γ-frequency subbands. A major limitation of this meta-analysis is the absence of a unified gold standard against which diagnostic accuracy measures are calculated. Some authors accepted DCS mapping results as the gold standard, while others considered fECoG to be more informative, as this method more accurately reflects the complex processes of cerebral cortex functioning and allows their assessment in dynamic conditions. For intraoperative fECoG application, analysis of clinical outcomes is complicated by postoperative tissue edema and neuroplasticity effects, which may influence both early and long-term clinical outcomes of cerebral cortex interventions.

Thus, fECoG is an informative and promising modality that requires validation, primarily for intraoperative application. To achieve this, a consensus on the gold standard is needed, which may include data on functional connectivity of the studied cerebral cortex regions.

Despite their high accuracy, a significant limitation of the aforementioned methods is the very fact of their invasiveness. While intraoperative brain stimulation is essentially a step in the neurosurgical procedure itself, the need for and protocol of extraoperative mapping must be considered when selecting electrode type and planning implantation. For example, a comparative meta-analysis of complications associated with depth versus subdural electrode implantation in patients with drug-resistant epilepsy showed that infectious complications develop significantly more often with subdural implantation (1.8% vs. 0.3% of cases), although the rates of neurological deficits and hemorrhage are approximately similar [24].

Noninvasive Mapping Methods

Functional Magnetic Resonance Imaging

Functional magnetic resonance imaging (fMRI) is an indirect method for visualizing areas of neuronal activity by assessing local hemodynamic response (changes in blood flow, more precisely, changes in blood chemistry) resulting from the phenomenon of neurovascular coupling. The history of fMRI began in 1990 when S. Ogawa and colleagues at Bell Laboratories proposed using blood-oxygenation-level-dependent (BOLD) contrast to study brain physiology with MRI. This contrast enables visualization of a minimal (approximately 2%) increase in MRI signal on gradient-echo sequences, caused by the washout of deoxyhemoglobin by oxyhemoglobin during activation of nearby neurons [25]. Numerous studies have demonstrated that such changes in MRI signal correlate quite well with the local electrical potential around a group of cells, reflecting postsynaptic activity [26]. Early fMRI studies used alternating blocks of activation and rest to identify regions responsible for performing specific tasks. Subsequently, following the work of B. Biswal et al., resting-state fMRI entered scientific and clinical practice — visualizing brain regions with correlated physiological low-frequency (0.01–0.08 Hz) fluctuations in BOLD signal during rest [27]. Because resting-state fMRI does not require specialized MRI-compatible equipment for task presentation or patient cooperation, and a single scanning session yields a dataset sufficient to map all functional systems — brain networks — it has currently replaced task-based fMRI in many scientific studies and partially in clinical practice [28].

In clinical practice, both methods coexist and are used for presurgical planning (to assess optimal surgical approach, predict surgical outcomes, evaluate neuroplastic processes and reorganization in the setting of brain tumors, and as a replacement for the Wada test [29]), as well as in neurorehabilitation to assess recovery potential and design rehabilitation programs, investigate functional reorganization, and identify targets for neuromodulation (e.g., mapping the dorsolateral prefrontal cortex for transcranial magnetic stimulation) (Fig. 2) [30].

 

Fig. 2. Examples of fMRI mapping.

A — primary sensorimotor cortex (SM1) of the hand in the healthy and affected hemispheres against peritumoral edema; B — dorsolateral prefrontal cortex (arrow) as a target for transcranial magnetic stimulation in a patient with moderate memory impairment; C — primary sensorimotor cortex of the foot and the corresponding sensorimotor network in a patient after stroke, shown in temporal dynamics.

 

Methodological limitations of fMRI do not yet allow it to fully replace invasive mapping methods. First, the study cannot be performed in patients for whom MRI is contraindicated (e.g., those with a pacemaker or MRI-incompatible implants). Second, the final outcome is influenced by the type of brain disorder (e.g., hemosiderin and postoperative metallic remnants in the bone or scalp may cause artifacts, while tumors or arteriovenous malformations may alter blood flow distribution, affecting detection of BOLD signal changes); patient cooperation (cognitive impairment may interfere with task performance, while sleep and emotional state may affect resting-state fMRI results); any patient movement during scanning; the relevance of the fMRI paradigm to the stated objectives; as well as technical characteristics — magnetic field strength of the scanner, parameters of fMRI sequences, quality control of images and postprocessing steps, the necessity of signal averaging, and interpretation of study results.

Most of the aforementioned factors can be mitigated; however, there are two substantial problems that currently cannot be successfully resolved. The first is the principle of MR signal acquisition, which affects the spatial accuracy of mapping. Activation in fMRI is an indirect visualization of changes in venous blood flow, where small parenchymal venules are typically located up to 1.5 mm from the site of neuronal activity, and larger draining veins up to 5 mm away. Thus, the region of presumed neuronal activity determined by fMRI may in fact be displaced from the true location of the pool of neurons of interest due to shifting toward large draining veins [31]. The second issue is the choice of statistical activation threshold. At present, there is no single optimal statistical threshold that can be used for fMRI assessment (Fig. 3), because the researcher must always seek a compromise between two types of error: false positives (noise being interpreted as activity) and false negatives (true activity remaining undetected). A stringent threshold (e.g., p < 0.001 adjusted for multiple comparisons) sharply reduces the likelihood of false positives, meaning the researcher can be confident that detected activity is real. However, such a threshold also significantly increases the likelihood of false-negative results — weak but biologically significant activity may be missed. This is particularly problematic for studies of complex cognitive processes or clinical populations, where effects may be less pronounced. Conversely, a low threshold (e.g., p < 0.01 without stringent adjustment) increases sensitivity, allowing detection of weaker effects, but the risk of false-positive results rises sharply. Due to the enormous number (hundreds of thousands) of voxels in the brain, even at p < 0.01 one could expect thousands of false detections.

 

Fig. 3. Example of activation area expansion (primary sensorimotor cortex during hand movement in the affected hemisphere (red) and healthy hemisphere (blue)) as the statistical threshold is lowered in fMRI data processing software.

The red arrow indicates the direction for the movement of the control used to lower the statistical threshold. As the threshold decreases, the activation area is observed to approach the tumor (asterisk).

 

As one potential solution, an approach using the peak voxel value for activation clusters has been proposed; however, a single such peak voxel cannot reliably represent a broader population of active neurons.

The relatively low temporal fMRI resolution, inherent to its methodology, should also be mentioned. In the most common clinical protocols, the repetition time — which determines the duration of data acquisition for a single brain volume unit — is 2–3 s. The widespread availability of high-field and ultra-high-field scanners and the implementation of parallel imaging techniques can reduce this parameter to 1 second or less (0.3–0.5 s), yet this remains far from the temporal resolution of electrophysiological methods, which reflect changes in brain function in near real time [32].

Despite the aforementioned limitations and considering the progress in technical development of scanners over their more than 30-year history, fMRI has strong potential for clinical application due to its noninvasiveness and high spatial resolution — and, in the case of resting-state fMRI, its ease of performance.

Navigated Transcranial Magnetic Stimulation

Navigated transcranial magnetic stimulation (nTMS) is a noninvasive method of brain stimulation based on the ability to excite cortical neurons by delivering brief magnetic pulses generated by a magnetic coil.

The emerging technical capabilities for nTMS mapping are associated with neuronavigation systems integrated with the magnetic stimulator [33]. These systems allow tracking and storing the locations of delivered stimuli based on individual structural MRI data of the subject, and subsequently exporting the resulting point sets for additional quantitative analysis or coregistration with mapping data from other modalities. In addition to the neuronavigation system, focality of stimulation is a necessary condition for nTMS mapping, which is determined by the geometry of the stimulating coil. Only figure-of-eight coils, which generate a field with a local maximum in the central region, meet these requirements [34].

nTMS mapping of the motor cortex involves sequential delivery of stimuli to the area of potential localization of the cortical representation of the target muscle, as well as to the adjacent premotor and primary somatosensory cortex. The set of points at which stimulation induces motor evoked potentials (MEPs) in a given muscle constitutes a map of its cortical representation [35].

Depending on the clinical objective, the motor nTMS FCCM protocol can vary significantly. Key variables include the choice of target muscle, coil orientation, use of a grid algorithm (when stimuli are delivered to cells of a predefined grid) and consequently grid size, the number of stimulus repetitions, muscle state, stimulation intensity, and others [36].

In clinical practice, the most significant area of application is presurgical FCCM in patients with tumors of eloquent areas (Fig. 4).

 

Fig. 4. TMS motor mapping in a patient with a space-occupying lesion of the right frontal lobe involving motor areas.

Motor mapping data are overlaid on the individual brain image in the ANT Neuro software. The white cross on the left image indicates the space-occupying lesion. Red points in the right image indicate sites where stimulation elicited motor evoked potentials with an amplitude > 50 µV; gray points indicate sites where stimulation did not elicit motor evoked potentials.

 

The primary goal of nTMS FCCM is to determine the spatial relationship between the location of the space-occupying lesion and the cortical muscle representation. The utility of nTMS motor FCCM is supported by increased extent and radicality of resection, as well as improved surgical clinical outcomes [36, 37]. The accuracy of TMS motor FCCM when compared with intraoperative DCS ranges from 0.4 to 14.8 mm [38]. Of clinical importance is the fact that nTMS motor FCCM retains its accuracy even in recurrent gliomas [39].

In addition to the motor cortex, nTMS also enables investigation of non-motor areas. In clinical practice, speech mapping is the most frequently used non-motor application. The technique involves delivering a train of repetitive magnetic pulses during the execution of a speech task with clinical assessment of the effect. Typically, a train of 5 pulses at a frequency of 5 Hz is delivered, i.e., stimulus duration of 1 second, with a delay of 0 ms between image presentation and stimulus delivery [40].

For speech paradigms in nTMS speech FCCM, naming of objects, actions, and pseudowords is most commonly used. Homologous regions in both hemispheres are stimulated, with each site stimulated at least three times in random order. A site is considered functionally significant if stimulation disrupts task performance in at least 2 out of 3 trials.

In 2017, a consensus protocol for nTMS speech FCCM was presented, recommending the following stimulation parameters: interstimulus interval between presentation of two consecutive images — 2500 ms; single image presentation time — 700 ms; stimulation intensity — 120% of the resting motor threshold of the hand muscle in the corresponding hemisphere [39]. However, from our own experience and published data, we conclude that parameters such as the number of pulses per train, their frequency, the delay between paradigm presentation and stimulation, and the stimulus tasks themselves may vary.

One of the disadvantages of nTMS speech FCCM is the subjectivity of speech error assessment, even by a speech specialist. Automated algorithms for detecting both errors and prolonged reaction time are being developed [41, 42]. In clinical practice, object naming is most commonly used, but evidence suggests that alternative tasks, such as pseudoword reading, may be more informative.

The method is well tolerated in young children. Despite the need for higher stimulation intensity and relatively limited speech abilities, complete bilateral maps can be obtained in 70% of children [43]. nTMS speech FCCM cannot serve as a full alternative to intraoperative DCS mapping, as it has significantly lower “disruptive” efficacy and a high negative predictive value — meaning that mapping of areas where stimulation did not cause speech disruption is most reliable. For “positive” areas, the reliability is low, always requiring intraoperative verification. In cases where awake surgery and prolonged fMRI studies are not feasible (e.g., in young children), nTMS speech FCCM may be the only available modality.

In addition to higher accuracy compared with fMRI, advantages of nTMS motor FCCM include the absence of any requirement for active patient participation, allowing mapping to be performed in patients with severe paresis, obtundation, or cognitive and visual impairments.

An important issue with serial nTMS mappings is the low reproducibility of map parameters [44]. The reasons for this may include both inherent method limitations and physiological intra-individual variability due to excessive variability of motor cortex neural connections, even in healthy individuals. Several approaches have been proposed to improve reproducibility, both by overcoming methodological limitations [45] and by reducing physiological variability — for example, using state-dependent nTMS [46].

The main prospects for TMS motor mapping include the development and refinement of mapping algorithms and protocols (adaptive algorithms, robotic devices, etc.), as well as the study of the causes of physiological variability, and the accumulation of large empirical datasets on the parameters of cortical representation maps in healthy individuals and in various diseases associated with motor system damage. This, in turn, will allow the identification of more accurate markers for differentiating normal from abnormal findings [47].

Stimuli for Functional Assessment of Non-Motor Cortical Areas

Unlike the primary motor cortex, mapping of structures involved in sensory perception, speech and motor processing, memory, and other complex functions requires patient interaction and clinical assessment of performance on specially selected tasks (stimuli) during FCCM. Among non-motor cortical regions, speech and sensory areas, as well as brain regions involved in memory, are of practical interest. In certain cases, mapping of other functions, typically related to a patient’s professional activities, may be necessary. Skills such as playing a musical instrument, chess, and multilingualism may also become objects of FCCM and require the preparation of appropriate tasks [48–50]. In recent years, there has been growing interest in techniques that identify social functions of the non-dominant hemisphere, such as facial expression recognition, visuospatial functioning, and social cognition [51–53]. The choice of stimuli for mapping depends not only on the anatomical location of the area being mapped but also on the technique and purpose of FCCM.

The first minimally invasive method for language lateralization was the selective intracarotid amobarbital test, proposed in the 1950s and named the Wada test after its developer [54]. Its primary objective is lateralization of language and memory. Since the test involves sequential temporary inactivation of one hemisphere via intra-arterial anesthetic to assess their functional contribution to memory and speech, the stimuli consist of a set of simple instructions, visual and verbal items that must fit within the relatively short duration of anesthetic effect [55, 56]. For language lateralization, assessing comprehension of simple instructions (serial counting, opening/closing eyes, tongue protrusion, smiling), naming of pictured objects, word and sentence repetition, and reading are sufficient. For assessing the role of the temporal lobes in memory, memory stimuli are presented to the patient after drug administration: pictured objects, three words spoken by the examiner, and reading of two sentences on a tablet computer screen [57].

Despite the utility of the Wada test for language lateralization and for predicting the risk of postoperative amnesia following temporal resections, it has several limitations that currently significantly restrict its application, namely, its relative invasiveness, the inability to precisely localize functions, the inability to assess non-language functions, and high dependence on the experience of the specialist interpreting the patient’s responses [58].

Since the goal of invasive non-motor FCCM using DCS is detailed localization of functional areas, the choice of stimuli becomes a carefully tailored tool for assessing cortical functions. Stimuli in non-motor DCS mapping have fairly high predictive accuracy and specificity, because when disruptions occur during task performance, they provide precise information about postoperative outcomes if the stimulated area is resected [59]. Simple stimuli with short articulation duration allow repeated testing of large cortical areas under the time constraints of a neurosurgical procedure, while their high functional specificity enables fine differentiation of adjacent cytoarchitectonic fields with high precision.

However, despite the numerous advantages of non-motor DCS mapping, there are several limitations, the foremost of which is the requirement for patient awakening and cooperation. Consequently, the level of wakefulness and stress during surgery, the ability to comprehend staff commands, and overall patient compliance affect the results and accuracy of FCCM. Continuous non-motor mapping, which requires intense mental activity on the part of the patient, leads to patient exhaustion and the risk of false-positive results [60]. Therefore, intraoperative non-motor FCCM using DCS requires special psychological preparation and patient selection at the preoperative stage, as well as the qualification of the specialist administering the stimuli and evaluating responses. The limited duration of awake surgery and the consequently simplified stimulus batteries do not allow assessment of the full spectrum of cognitive functions beyond speech [61].

The use of different stimulus batteries during extraoperative DCS mapping is limited to the area covered by implanted subdural or depth electrodes and is therefore typically dictated by the clinical objectives of the invasive EEG. Most stimuli for extraoperative non-motor FCCM are similar to those used during awake surgery. However, extraoperative FCCM allows deeper investigation of mental processes and neural network functioning. This is due to the larger number of patients with comprehensive coverage of brain areas by subdural electrodes and the depth of electrode placement for stereo-EEG.

The future direction for the development of stimulus batteries for FCCM lies not in unification, but in the integration of multidirectional stimulus paradigms, where each battery progressively addresses specific diagnostic questions. Skillful combination of different stimuli allows a departure from the simplistic search for functional points and moves toward creating an individualized, integrated functional organization of the brain. With this approach, stimulus selection and application become key — a means to minimize functional damage and make informed surgical decisions.

Discussion

Despite its long history, several fundamental and clinical issues regarding FCCM using DCS remain unresolved. The development of methods based on analysis of metabolic changes (fMRI, fECoG) and their frequent discordance with DCS mapping raises questions about the accuracy of the latter. There is still no complete understanding of the stimulation process itself or the rules governing the spread of electrical current between the anode and cathode in the heterogeneous tissues of the brain (gray and white matter). There are no unified protocols for stimulation parameters or for the selection of tasks for cognitive FCCM, particularly in the pediatric population. No studies have examined the influence of antiepileptic drugs on cortical excitability and, consequently, on current intensity thresholds for DCS. Investigating complex, distributed network cognitive functions (e.g., speech and memory) through brief interruption of only a small cortical area by an electrical or magnetic pulse limits confidence in the reliability of this mapping approach. Existing electrical stimulation parameters are not physiological, as evidenced, among other things, by the unnatural sensations reported by subjects with implanted bidirectional brain–computer interfaces [62].

fMRI and nTMS are noninvasive methods for preoperative functional mapping that can be applied in specific patient populations, including those with limitations for invasive and intraoperative techniques. Functional electrocorticography is a promising invasive mapping modality that does not use electrical current and enables assessment of complex dynamic processes in speech mapping. Criticism of electrical stimulation techniques, supported by everyday clinical practice, necessitates further refinement of other FCCM methods and possibly the abandonment of DCS as the gold standard for assessing their accuracy. The only true reference method wouldbe the presence of a permanent postoperative focal neurological deficit following resection of a mapped functional area; however, medical ethics does not permit the use of such a marker as a substitute for DCS. In the development of activation methods (fMRI, fECoG) for non-motor FCCM, there is another limitation that may explain their relatively low sensitivity compared with DCS: existing stimulus batteries were developed for stimulation-based (DCS, nTMS) mapping, and their transfer to a method with a different operating principle requires separate validation under new conditions. Thus, all of the methods presented require further standardization, development of stimulus materials, and consensus on the gold standard for comparing results.

At the end of 2025, a scientific conference “Functional Mapping of the Cerebral Cortex: Practical Approaches and New Technologies” was held in Moscow, bringing together researchers studying cerebral cortex functions. The main topic of lively discussion was clinical practice and the limitations of applicability of existing mapping methods among neuro-specialists: neurologists, clinical neurophysiologists, neurorehabilitation specialists, neuropsychologists, and neurosurgeons involved in the treatment and rehabilitation of patients with drug-resistant epilepsy and brain tumors. The main points of this discussion are presented in this work. The conference resulted in a consensus that extraoperative DCS should be adopted as the basis for clinical application. fMRI and nTMS are complementary methods with certain limitations, while fECoG is at the stage of clinical validation and emerging widespread use. Additionally, an important objective is the development of multidirectional stimulus materials based on current knowledge of cognitive functions, taking into account the specific features of each functional mapping technique.

Conclusion

Both invasive and noninvasive FCCM are widely used in clinical practice. Despite numerous limitations, DCS remains the most accurate and reproducible method; however, for activation methods, the adoption of a unified reference method is a cornerstone for their further study and implementation into clinical practice.

The current arsenal of cerebral cortex functional mapping methods is complementary, and the choice among them is determined by the specific clinical task. A promising direction is not the search for a single universal method, but rather the convergence of technologies and the creation of multimodal integrative platforms. Such technological systems should combine the strengths of invasive and noninvasive approaches, as well as utilize standardized, adaptive stimulus paradigms and data processing algorithms. This will enable a transition from point localization of functions to the construction of individualized maps of functional networks and the prediction of outcomes.

About the authors

Mikhail V. Sinkin

Sklifosovsky Research Institute of Emergency Medicine; Pirogov Russian National Research Medical University

Author for correspondence.
Email: mvsinkin@gmail.com
ORCID iD: 0000-0001-5026-0060

Dr. Sci. (Med.), leading researcher, Emergency neurosurgery department; Head, Department of medical neurotechnology, Institute of Neuroscience and Neurotechnology

Russian Federation, Moscow; Moscow

Elena I. Kremneva

Russian Center of Neurology and Neurosciences

Email: kremneva@neurology.ru
ORCID iD: 0000-0001-9396-6063

Dr. Sci. (Med.), senior researcher, Radiology department, Institute of Clinical and Preventive Neurology

Russian Federation, Moscow

Alexandra G. Poydasheva

Russian Center of Neurology and Neurosciences

Email: alexandra.poydasheva@gmail.com
ORCID iD: 0000-0003-1841-1177

Cand. Sci. (Med.), researcher, Noninvasive neuromodulation group, Institute of Neurorehabilitation and Rehabilitation Medicine

Russian Federation, Moscow

Anastasia A. Skalnaya

Sklifosovsky Research Institute of Emergency Medicine; Pirogov National Medical and Surgical Center

Email: mvsinkin@gmail.com
ORCID iD: 0000-0001-5938-9194

neurologist, Neurosurgical department; junior researcher, Emergency neurosurgery department

Russian Federation, Moscow; Moscow

Tatiana A. Salomatina

Almazov National Medical Research Center

Email: mvsinkin@gmail.com
ORCID iD: 0000-0001-5106-0589

medical psychologist, Russian Research Neurosurgical Institute named after Professor A.L. Polenov

Russian Federation, Saint-Petersburg

References

Supplementary files