**1. Introduction**

Navigated transcranial magnetic stimulation (nTMS) combines the use of neuronavigation with TMS to target neurostimulation inductively to the brain cortex, utilizing views of the brain anatomy with sub-centimeter precision and enabling tracking of the coil during stimulation (e.g., using an infrared (IR) tracking system combined with the stimulator; Figure 1) [1]. For over a decade, nTMS has been used in diagnostic setups (e.g., to perform non-invasive clinical mapping of motor or other brain functions) or for therapeutic purpose (e.g., to treat major depression or other psychiatric diseases as well as chronic pain). Preceding nTMS as we know it today, early approaches conducted mapping based on anatomical

**Citation:** Sollmann, N.; Krieg, S.M.; Säisänen, L.; Julkunen, P. Mapping of Motor Function with Neuronavigated Transcranial Magnetic Stimulation: A Review on Clinical Application in Brain Tumors and Methods for Ensuring Feasible Accuracy. *Brain Sci.* **2021**, *11*, 897. https://doi.org/ 10.3390/brainsci11070897

Academic Editor: Lorenzo Rocchi

Received: 4 June 2021 Accepted: 2 July 2021 Published: 7 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

landmarks placed over the scalp or used frameless stereotactic systems to localize and track stimulations [2–4]. However, such approaches seemed limited for clinical use where high precision and reliability is warranted in relation to individual brain anatomy that can be both spatially and functionally deranged due to pathology. Technological advancements have enabled the estimation of the spatial extent and geometry of the electric field (EF) induced by stimulation [1,5–7]. Specifically, the development of nTMS using EF-based neuronavigation acknowledges that the EF induced by stimulation strongly depends on several factors such as skull thickness and coil tilting, amongst others. In this context, accurate and efficient modelling of the EF is essential to pinpoint the impact of stimulation and to understand its effects [1,5–7]. Since nTMS using such EF-based neuronavigation is currently considered the most accurate method for targeting stimulation, this article focusses on this technique.

**Figure 1.** Setup for nTMS mapping. (**A**) Hardware of the nTMS system, including an infrared camera, EMG amplifier, neuronavigation monitor, and the stimulator with a cooling unit as the central components. (**B**) Patient positioning during mapping procedures, requiring initial co-registration of the patient's head with cranial MRI (using a head tracker attached to the patient's forehead) to be able to track the stimulating coil (equipped with infrared trackers) in relation to individual brain anatomy. (**C**) Navigation view during the mapping procedure, showing the stimulating coil (*yellow marker*) with its orientation (*red arrow*) and modelled EF distribution.

Overall, interest towards combining neuronavigation with TMS has experienced more and more interest in the research community as indicated by the increasing trend in published studies (Figure 2). From a clinical perspective, nTMS has been gaining importance especially as a tool for preoperative mapping used for planning and intraoperative resection guidance in patients harboring functionally eloquent brain neoplasms [1,8–11]. As of today, mapping of motor function has become the mainstay of nTMS, making so-called nTMS motor mapping in patients with motor-eloquent lesions an ideal use case for the technique. In modern neuro-oncological surgery, achieving an optimal balance between the extent of resection (EOR) of a brain tumor and the individual functional status of the patient is the major principle for surgical resection [12]. Prognostically, a safely performed maximized tumor resection is of utmost importance as incomplete resection is correlated to lower survival rates and quality of life for patients with malignant glioma as the prominent entity of intra-axial brain tumors [13–18]. To maximize the EOR while keeping surgeryrelated decline of function (e.g., paresis in the context of the motor system) to the lowest level achievable, intraoperative direct electrical stimulation (DES) is performed as the goldstandard method for assessing subcortical and cortical functional representations [19–23]. Outside of the operating theater, nTMS motor mapping and later-developed nTMS-based tractography can be performed for this purpose. As the major field of current application,

**Figure 2.** Number of articles on nTMS published annually between 2002 and 2020 as indexed in PubMed (on the 21st of May in 2021). The search query in PubMed was defined as follows: ("navigated transcranial magnetic stimulation" OR "navigated TMS" OR "neuronavigated TMS" OR "neuronavigated transcranial magnetic stimulation").

From a methodological perspective, the nTMS parameters selected for motor mapping, such as the stimulation intensity (SI), number of applied stimuli, and the density of the applied stimulation grid, affect the outcome and accuracy of the motor mapping (Figure 3) [24–27]. Specifically, optimal selection becomes especially important when quantitative parameters are drawn from the motor maps, e.g., to characterize the location of motor function or to define its extent on the cortex. Quantitative mapping refers to those quantitative parameters derived from nTMS motor mapping, based on the recorded responses and location information related to the responses. The quantitative parameters in nTMS motor mapping include the locations of the motor hotspot and center of gravity (CoG), and area and volume of the motor map, which are all based on recorded amplitudes of motor-evoked potentials (MEPs) elicited by stimulation (Figure 3) [24–27]. The motor hotspot location is often considered as the location of the maximum MEP amplitude response [28,29]. It not only characterizes the major location of the motor representation, but is also used as the location where the resting motor threshold (rMT) is determined and around which the motor map extends. The CoG is defined as the amplitude-weighted location representing the location in the motor representation area where the center of motor activation lies [28,30,31]. The volume of the motor map is usually calculated by summing the MEP amplitudes within a motor map, e.g., within a stimulus grid [30–33]. The area of the motor map on the other hand binarizes the MEP responses to positive and negative responses and represents the area covered by positive responses (i.e., spots where stimulation elicited an MEP with a certain amplitude) [27,34,35]. Importantly, wrong selection or wrong interpretation of these quantitative parameters affects the confidence in the motor map and, in the worst case, could lead to imprecise motor maps that may have harmful consequences to the patient when used for preoperative planning and intraoperative resection guidance in neuro-oncological surgery. The technical solutions and refinements realized in the nTMS systems by the manufacturers are also important for the overall performance of motor mapping, albeit representative comparisons between different manufacturer-specific approaches have not yet been made in the literature. Given the crucial role quantitative parameters play for nTMS motor mapping, the second part of this article demonstrates the significance of certain parameters used in motor mapping to

apply quantitative measures on the resulting individual motor maps. Finally, we speculate on the future of nTMS motor mapping by the potential in advancing from current trends towards purely quantitative motor mapping, which would have direct clinical impact. Relevant studies were identified by PubMed search (http://www.ncbi.nlm.nih.gov/pubmed; accessed on 21 May 2021).

**Figure 3.** Overview of the main steps during the nTMS mapping procedure, starting with upload of the MRI data to the nTMS system and resulting in the generation of the individual motor map that can be further used for nTMS-based fiber tracking and other interdisciplinary or multi-modal applications. In addition, the main quantitative parameters that can be extracted from the single steps are shown (including the location of the motor hotspot, SI, number of stimuli, density of the stimulation grid, CoG, area, and volume of the motor map).

#### **2. Clinical Application of nTMS for Mapping Motor Function**

*2.1. Feasibility, Reliability, and Comparison with Other Methods*

The gold standard for functional brain mapping is defined by DES. Thus, any other technique for mapping purposes needs to be validated against it. First case reports in patients with intracranial metastasis (MET), meningioma (MEN), and low-grade glioma (LGG) have already indicated good agreement between preoperative motor mapping by nTMS and DES [32–34]. Specifically, in an early series of 20 patients with different entities of intracranial neoplasms, the motor hotspot was located on the same gyrus for nTMS and DES mappings in all enrolled cases, with distances of motor hotspots between techniques amounting to values <1 cm for specific cortical muscle representations [35]. In this regard, Table 1 provides an overview of studies on reliability of nTMS motor mapping and comparison with other methods, mainly including DES.

The initial impression of high concordance between motor maps provided by preoperatively acquired nTMS motor mapping and those derived from DES was confirmed by subsequent studies that aimed to enroll larger patient cohorts with different entities of

brain lesions and performed comparisons with a third method. Specifically, comparative analyses were achieved for nTMS and task-based functional magnetic resonance imaging (fMRI) versus DES mappings since task-based fMRI appeared as a widely used alternative to nTMS, with all major studies consistently revealing closer distances for localizations of cortical motor function between preoperative nTMS and DES compared with task-based fMRI activation maps [36–41]. Exemplarily, in a series of 26 patients, distances between motor hotspots of nTMS and DES amounted to 4.4 ± 3.4 mm on average [38]. However, when comparing distances between activation in task-based fMRI and nTMS motor hotspots, larger distances of 9.8 ± 8.5 mm for upper extremity (uE) and 14.7 ± 12.4 mm for lower extremity (lE) muscle representations were observed [38]. More recently, considering motor representations of the uE, lE, and tongue in 36 patients with intracranial tumors, there were significantly smaller Euclidean distances as well as a better spatial overlap between DES and nTMS compared with task-based fMRI [41]. In this regard, one study suggested that nTMS motor mapping is as accurate in recurrent gliomas as it seems to be prior to the initial surgery, opening up the possibility to reuse nTMS motor mapping in the individual course of disease [39]. Similar to results for nTMS motor mapping versus fMRI, in an investigation comparing the accuracy of nTMS with magnetoencephalography (MEG) and DES mapping, distances between motor hotspots of nTMS and DES mappings were significantly smaller than those between the activation in MEG and nTMS motor hotspots [42].

The implementation of nTMS motor mapping confirmed the expected anatomy in 22%, added awareness of high-risk functional areas in 27%, modified the approach in 16%, changed the planned EOR in 8%, and changed the surgical indication in 3% of patients among a cohort of 73 patients with different entities of brain neoplasms [43]. Similarly, another study concluded that nTMS motor mapping enabled an exact localization of the motor cortex in 88.2%, provided the neurosurgeon with new unexpected information about functional anatomy in 70.6%, and facilitated a modification of the surgical approach to spare the motor cortex from damage in 29.4% of patients [44]. Further, the availability of nTMS data improved the neurosurgeons' confidence in identifying the motor strip or central region [38,40]. Importantly, valuable motor mapping data from nTMS may be generated irrespective of the distinct experience level of the operator (after a certain interval of training), with average intra-examiner distances between CoGs for an expert investigator amounting to 4.4 mm and 4.9 mm for the expert versus novice investigator, with comparable values in the investigated healthy subjects as well as in patients with brain tumors [45].

Regarding potential variability of MEP latencies or the SI applied during nTMS motor mapping as a function of the rMT, it has been revealed in large cohorts of patients with motor-eloquent tumors that sex and antiepileptic drug (AED) intake, amongst other factors, may considerably contribute to MEP latency variability as well as variability of the determined rMT [46,47]. Interestingly, also the tumor grading according to the World Health Organization (WHO) scheme may be held accountable for a certain degree of variability introduced to nTMS motor mappings because MEP latencies of lE muscles increased with the WHO grading, and correlations between the increase in WHO grading and a decreased rMT were observed for lE muscles [48]. From a more methodological perspective, efforts have been made to expand accurate nTMS-based motor mapping beyond the limits of the supposed primary motor cortex, and to increase robustness by introducing altered threshold values for MEP amplitudes [49,50]. Future research will determine the distinct relevance of such influencing factors for motor mapping by nTMS in patients with brain tumors, in whom such potential confounders of motor mapping results are still largely neglected in the clinical routine workflow.

**Table 1.** Feasibility, reliability, and comparison with other methods. This table outlines the studies published on feasibility and reliability of motor mapping by nTMS (using an electric-field-navigated system) in patients harboring brain neoplasms. Furthermore, studies comparing nTMS motor mapping with other modalities (intraoperative DES, preoperative fMRI, and preoperative MEG) are included.



**Table1.***Cont.*


**Table1.***Cont.*


**Table1.***Cont.*

Abbreviations: nTMS—navigated transcranial magnetic stimulation; LGG—low-grade glioma; HGG—high-grade glioma; MET—metastasis; MEN—meningioma; rMT—resting motor threshold; aMT—active motor threshold; uE—upper extremity; lE—lower extremity; DES—direct electrical stimulation; fMRI—functional magnetic resonance imaging; APB—abductor pollicis brevis; TA—tibialis anterior; EOR—extent of resection; MEG—magnetoencephalography; CoG—center of gravity; ADM—abductor digiti minimi; FCR—flexor carpi radialis; MEP—motor-evoked potential; AED—antiepileptic drug; SI—stimulation intensity; WHO—World Health Organization.

#### *2.2. Fiber Tractography*

The motor-positive nTMS points can be used for seeding to obtain regions of interest (ROIs) for subsequent delineation of the corticospinal tract (CST) in the context of fiber tracking. The combination of nTMS motor maps at the cortical surface with diffusion MRI (dMRI) and tractography, i.e., the delineation of the primary motor cortex and its subcortical connections to the peripheral nervous system with visualization of the course of the CST, can provide a more complete picture than one of the techniques alone (Figure 4). In this regard, Table 2 outlines studies using data from nTMS motor mapping for tractography of the CST in patients with different kinds of brain neoplasms.

In a first tractography study using nTMS data, 30 patients harboring motor-eloquent brain tumors underwent presurgical nTMS motor mapping followed by nTMS-based tractography, using the nTMS motor-positive spots and the ipsilateral cerebral peduncle as ROIs [51]. Compared with the conventionally used approach with manual delineation of the suspected primary motor cortex and the ipsilateral cerebral peduncle as ROIs, the novel setup led to a lower number of fibers displayed, a reduced volume of the CST, and, most importantly, lower fractions of aberrant tracts most likely not belonging to the CST [51]. The approach of nTMS-based tractography of the CST was subsequently refined by investigating individually adapted adjustments for the fractional anisotropy (FA) that had to be defined for the deterministic tractography algorithm: the FA was increased stepwise until no fibers were displayed, followed by reducing the FA value by 0.01, thus delineating only a thin fiber course; the obtained FA value was defined as 100% FA threshold (FAT), and nTMS-based tractography was then carried out with 50% and 75% FAT with motor-positive nTMS points and the manually delineated internal capsule or brainstem as ROIs [52]. This method influenced the surgical strategy in 46% of patients, in contrast to conventional tractography without nTMS data for ROI generation where an impact was only observed for 22% of patients [52].

Furthermore, a study among 20 patients with different entities of brain neoplasms achieved detailed somatotopic CST reconstruction derived from nTMS motor maps combined with dMRI for uE, lE, and face muscle representations, with a decreased number of fibers and a greater overlap between the motor cortex and the cortical end-region of the CST when compared with conventional tractography with only anatomical seeding [53]. Of note, the obtained CST course as well as the somatotopic organization were confirmed by DES mapping [53]. In another study on somatotopic reconstruction of the CST considering parts subserving uE, lE, and face muscles, a higher fraction of plausible fibers was observed for seeding at the anterior inferior pontine level when compared with seeding at the internal capsule, combined with nTMS-based seeding at the cortical level [54]. When setting somatotopic nTMS-based tractography in contrast to fMRI-based seeding, a higher plausibility was observed for the nTMS-based approach, and fMRI-originated tracts presented with a more posterior course relative to the nTMS-based reconstruction of tracts [55]. Recently, in a comprehensive study systematically comparing different setups for tractography (deterministic and probabilistic algorithms with variable ROI definitions) and correlating tractography with DES mapping and fMRI findings in 11 adult patients, highest accuracy of tractography was achieved when using seeding with a manually generated mask enclosing the precentral gyrus, but none of the applied setups showed clear superiority and nTMS- or fMRI-based tractography differed only slightly [56]. Yet, probabilistic tracking resulted in an optimized correlation with DES mapping when compared with the more commonly used deterministic tractography algorithm [56]. Upcoming work is needed to further investigate optimal settings and algorithms in representative cohort sizes to achieve results of CST tractography as close as possible to DES mapping results in order to assure accuracy and reliability of nTMS-based tractography.

**Table 2.** Fiber tractography. This table presents the studies published on tractography of the CST primarily using motor mapping by nTMS (using an electric-field-navigated system) for seeding in patients harboring brain neoplasms.



**Table 2.** *Cont.*

 Abbreviations: nTMS—navigated transcranial magnetic stimulation; LGG—low-grade glioma; HGG—high-grade glioma; MET—metastasis; MEN—meningioma; CST—corticospinal tract; dMRI—diffusion magnetic resonance imaging; ROI—region of interest; FA—fractional anisotropy; FAT—fractional anisotropy threshold; FL—fiber length; fMRI—functional magnetic resonance imaging; uE—upper extremity; lE—lower extremity; DES—direct electrical stimulation; SI—stimulation intensity.

**Figure 4.** Exemplary patient case (right-hemispheric glioma in a 56-year-old male patient) for illustration of CST reconstruction using tractography based on motor maps derived from motor mapping with nTMS. (**A**) Motor map with binarization into motor-positive (*white*) and motor-negative (*grey*) stimulation points. (**B**) Tractography of the CST (*orange*) based on a ROI constituted of motor-positive nTMS points (*green*). (**C**) Fusion of T1-weighted imaging and tractography results (axial plane). (**D**) Fusion of T1-weighted imaging and tractography results (sagittal plane).
