The influence of sentence length on the activation of the language network: a functional magnetic resonance imaging study
by Manuel Unnerstall
Date of Examination:2025-10-30
Date of issue:2025-10-02
Advisor:PD Dr. Nicole Neef
Referee:PD Dr. Nicole Neef
Referee:PD Dr. Vesna Malinova
Referee:Prof. Dr. Martin Sommer
Persistent Address:
http://dx.doi.org/10.53846/goediss-11529
Files in this item
Name:Unnerstall_Manuel_eDiss.pdf
Size:7.67Mb
Format:PDF
These files are not accessible until the state exam has been completed. !Es liegt auch ein Embargo vor!
Abstract
English
Understanding and processing language is one of the most crucial abilities in our everyday life. This ability is at risk when certain neurosurgical interventions involving supratentorial tissue alterations are necessary. The language network is subject to lateralization and distributed across different cortical regions. Interindividual variability affects the exact localization and the extent of activation. Preoperative mapping of brain structure to function can be achieved through functional magnetic resonance imaging (fMRI). This method requires the use of stimuli to activate the language network during the scan. A paradigm currently used at the Medical University Center Göttingen involves the presentation of four-word sentences, shown one word at a time. This approach keeps the demands on the verbal working memory comparable to those in the literature from which the method originates. This dissertation examines visually presented German four-, eight-, and twelve-word sentences as stimuli to compare the resulting effect sizes. Ten healthy, neurotypical participants (mean age = 42 years, min = 24 years, max = 61 years, 7 women) each completed two runs of functional magnetic resonance imaging, during which all three conditions were implemented block-wise in symmetrical, pseudorandomized order. The blood-oxygen-level dependent effect size, measured as percent signal change, were assessed within regions of interest that were functionally defined using data from the respective other run. T-tests were applied for each subject individually to determine whether longer sentences correlated with significantly greater mean effect sizes across the language network. Robust activation of the language network was achieved in individual participants. In eight out of ten cases, the eight-word and twelve-word sentences were associated with significantly higher mean effect sizes than the four-word sentences. Additionally, behavioral data from probe decision tasks support the recommendation of the eight- and twelve-word conditions over the four-word condition. Evidence suggests that the improved efficiency of longer sentences should be utilized in clinical application. The alternative stimuli can contribute to preoperative diagnostics with shorter scanning times and stronger activation of the language network due to higher semantic density. As a clinical example, the case of this department’s first patient who received a preoperative fMRI scan using the eight-word paradigm instead of the four-word paradigm is discussed. The duration of word and pseudoword presentation had to be adapted, and robust activation of the language network was again achieved. Overall, this study enables us to choose from different sentence lengths for stimulation based on evidence and direct, controlled comparison of their resulting effect sizes. In general, the use of functional magnetic resonance imaging, which originates in basic neuroscientific research, is expected to become more frequent in clinical practice. As a result, the need for advanced automation arises when collecting and analyzing raw data for clinical routine measurements rather than research.
Keywords: language network; language lateralization; fMRI; neuroradiology
