gms | German Medical Science

71. Jahrestagung der Deutschen Gesellschaft für Neurochirurgie (DGNC)
9. Joint Meeting mit der Japanischen Gesellschaft für Neurochirurgie

Deutsche Gesellschaft für Neurochirurgie (DGNC) e. V.

21.06. - 24.06.2020

TMS language mapping revisited – analysis of 90 patients in MNI space and machine learning classification

TMS Sprachmapping – Analyse von 90 Patienten im MNI-Raum mit anschließender Klassifizierung durch maschinelles Lernen

Meeting Abstract

  • presenting/speaker Ziqian Wang - Charité – Universitätsmedizin Berlin, Klinik für Neurochirurgie, Berlin, Deutschland
  • presenting/speaker Lucius Fekonja - Charité – Universitätsmedizin Berlin, Klinik für Neurochirurgie, Berlin, Deutschland
  • Felix Dreyer - Freie Universität Berlin, Brain Language Laboratory, Berlin, Deutschland
  • Peter Vajkoczy - Charité – Universitätsmedizin Berlin, Klinik für Neurochirurgie, Berlin, Deutschland
  • Thomas Picht - Charité – Universitätsmedizin Berlin, Klinik für Neurochirurgie, Berlin, Deutschland

Deutsche Gesellschaft für Neurochirurgie. 71. Jahrestagung der Deutschen Gesellschaft für Neurochirurgie (DGNC), 9. Joint Meeting mit der Japanischen Gesellschaft für Neurochirurgie. sine loco [digital], 21.-24.06.2020. Düsseldorf: German Medical Science GMS Publishing House; 2020. DocP080

doi: 10.3205/20dgnc367, urn:nbn:de:0183-20dgnc3677

Published: June 26, 2020

© 2020 Wang et al.
This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 License. See license information at http://creativecommons.org/licenses/by/4.0/.


Outline

Text

Objective: Non-invasive repetitive navigated transcranial magnetic stimulation (rnTMS) is increasingly used for preoperative cortical language mapping. While in recent years knowledge about the healthy human language network has constantly improved, the effect of language eloquent brain tumors on the language network remains largely unclear. Several studies investigated tumor-induced neuro-reorganization, but the reorganization pattern still needs to be analyzed. This study aims to examine the language network in regard to the impact of pathology by applying group analysis methods.

Methods: We retrospectively reviewed a cohort of 90 right-handed patients with left perisylvian WHO grade II-IV gliomas. All patients underwent preoperative rnTMS-language- mapping. The patients were classified into an aphasic and non-aphasic group. All TMS spots were registered from individual space into MNI space and parcellated using the automated anatomical labeling (AAL) template to obtain the error rate (ER) of each anatomical volume of interest (AVOI). Subsequently, univariate statistical analysis was performed for each ER of AVOI and biometric data between the two groups. The significant results were fed as features, e.g. input variables, into the support-vector machine (SVM), a supervised machine learning model,to classify aphasic and non-aphasic patients.

Results: 30 of 90 (33.3%) patients suffered from aphasia. Univariate analysis revealed 11 perisylvian AVOIs’ ERs (8 left, 3 right hemispheric) that were significantly higher in the aphasic than non-aphasic group (p < 0.05), depicting a broad, bihemispheric language network. After feeding the significant AVOIs into the SVM model, it showed that additional to age (w = 2.95), the ERs of right Frontal_Inf_Tri (w = 2.06) and left SupraMarginal (w = 2.05) and Parietal_Inf (w= 1.80) contributed more than other features to the model. The model’s sensitivity was 89.7%, the specificity was 82.0%, the overall accuracy was 81.1% and AUC was 88.7%.

Conclusion: SVM based group analysis revealed a distinct pattern of perisylvian language-network reorganization with especially the right inferior frontal gyrus showing a significantly higher ER in aphasic patients than in non-aphasic patients. Moreover, the model demonstrated that aphasic patients were significantly older than non-aphasic patients, indicating a reduced potential for language reorganization in elderly patients.