*2.4. Cranial Magnetic Resonance Imaging*

Imaging was performed on a 3-Tesla MRI scanner using a 32-channel head coil (Achieva dStream or Ingenia; Philips Healthcare, Best, The Netherlands). As part of a standardized, multi-sequence imaging protocol for brain tumors, a three-dimensional fluid attenuated inversion recovery (FLAIR) sequence (repetition time (TR)/echo time (TE): 4800/277 ms, 1 mm<sup>3</sup> isovoxel covering the whole head), three-dimensional T1-weighted gradient echo sequence (TR/TE: 9/4 ms, 1 mm<sup>3</sup> isovoxel covering the

whole head) without and with application of an intravenous contrast agent (Dotagraf, Jenapharm GmbH & Co. KG, Jena, Germany), and a DTI sequence (TR/TE 5000/78 ms, voxel size of 2 <sup>×</sup> <sup>2</sup> <sup>×</sup> 2 mm<sup>3</sup> , 32 diffusion gradient directions) were acquired preoperatively and were used for nTMS language mapping (contrast-enhanced T1-weighted sequences) and nTMS-based DTI FT (contrast-enhanced T1-weighted, DTI, and FLAIR sequences).

Postoperative (within the first 48 h after surgery) and follow-up imaging were performed using the same imaging protocol. In postoperative MRI, special attention was paid to the assessment of residual tumor tissue or achieved gross total resection (GTR) as well as perioperative bleeding or ischemia.

#### *2.5. Language Mapping by Navigated Transcranial Magnetic Stimulation*

#### 2.5.1. Mapping Procedure

Preoperative language mapping by nTMS of the tumor-affected left hemisphere (LH) was carried out during the days before scheduled surgery for tumor resection using a Nexstim eXimia NBS system (version 4.3; Nexstim Plc, Helsinki, Finland) [11,14,15,39,40]. Language mappings were conducted in German as the native language of all enrolled patients.

After initial co-registration of the patient's head and the contrast-enhanced T1-weighted image dataset, pictures of common objects were presented to the patient in the context of two baseline trials (presentation of objects without simultaneous stimulation) [13,40–44]. The objects shown were part of a standardized object-naming task, and the purpose of baseline assessment was to systematically discard objects that did not elicit a quick and fluent response and to familiarize the patients with the task and setup. The remaining stack of objects that were named correctly and fluently was then shown under stimulation of up to 46 target points that had been placed on the LH, with each point being stimulated six times in total (stimulation intensity: 100% of the individual resting motor threshold, stimulation frequency: 5 Hz/5 pulses) [13,40,42,43].

The 46 target points for stimulation were tagged with close respect to the cortical parcellation system (CPS), thus covering almost the whole LH except for polar, occipital, and inferior temporal regions, similar to previous studies on nTMS language mapping (Figure 1) [13,42,43]. Exclusion was due to considerable muscle activation and/or discomfort that can be observed following stimulation of these particular regions. Furthermore, in patients showing large tumor masses, the number of 46 target points for stimulation needed to be reduced according to individual cortical architecture given destructions and derangements caused by such lesions. The picture-to-trigger interval during stimulation was set as 0 ms by default [45,46]. The patients' performance during object naming was video-recorded for later evaluation to detect and categorize errors elicited by targeted nTMS [41,44].

#### 2.5.2. Mapping Evaluation

Video data of the nTMS language mappings were systematically searched for naming errors of different categories after the mapping procedure. No responses, performance errors, hesitations, neologisms, phonological paraphasias, and semantic paraphasias were defined and considered during evaluation [13,41–44]. The stimulation points that elicited any naming error except hesitations (no responses, performance errors, neologisms, phonological paraphasias, and semantic paraphasias together) were considered as language-positive nTMS points. Correspondingly, stimulation points that did not elicit a naming error were defined as language-negative nTMS points.

Per patient, two export datasets were then generated, which included the language-positive nTMS points of the LH and the language-negative nTMS points of the LH, respectively. Each export file was in Digital Imaging and Communications in Medicine (DICOM) format and included the nose and ears as anatomical landmarks together with the stimulation points. Furthermore, in each patient, CPS regions were counted as language-positive CPS regions when they contained at least one language-positive nTMS point, while CPS regions without language-positive nTMS points were considered as language-negative CPS regions. CPS regions that were not stimulated (polar, occipital, and inferior temporal regions) were not considered. The total number of language-positive CPS regions was recorded as Np, and the number of language-negative CPS regions as Nn.

**Figure 1.** Cortical parcellation system (CPS). The tumor-affected left hemisphere (LH) was mapped by navigated transcranial magnetic stimulation (nTMS) with respect to 46 target points that were placed in relation to the CPS (reduced numbers of target points were accepted in patients with large tumor masses, which hampered placement of the total amount of 46 target points). The points were stimulated six times in total each, with a stimulation intensity of 100% of the individual resting motor threshold and a stimulation frequency of 5 Hz/5 pulses. Numbers in circles schematically represent the stimulation targets on a standardized brain template, red dashes mark regions that were not subject to nTMS.
