gms | German Medical Science

Objective: Preoperative functional imaging modalities can be readily used to identify sub-/cortical regions involved in language processing, but fail to point out regions at risk regarding permanent deficits. To overcome this limitation, is particularly important to tailor surgical procedures involving networks with a high potential of functional reorganization, such as the language network. Recently, several groups reported the left temporo-parieto-occipital junction (TPO-L) and connected tracts (i.a., temporo-parietal sections of the arcuate fasciculus and the superior longitudinal fasciculus), as well as the inferior fronto-occipital fasciculus (IFOF) to be highly correlated with permanent language deficits. We here set out to identify a standard procedure for multi-task fMRI-informed tractography, optimized to depict the aforementioned structures at risk.

Methods: 24 healthy subjects (f =11, mean age= 24, right-handed, native German speakers) performed three fMRI tasks in German language: (1 - PN) picture naming, (2 - SC) simple auditory sentence comprehension (SC), (3 - SD) complex auditory sentence comprehension. Task 2 (SC) and 3 (SD) included a semantic decision task, i.e., selecting one semantically related object drawing out of three options via button-press. Moreover, SD included a paralleled nonsense sentence modality (ns-SD), in addition to meaningful sentences (1:4). Both SC and SD were controlled by a reverse speech modality. The different language tasks are compared to elaborate the optimal combination of tasks (contrasts and statistical thresholds) to define starting/ending regions of interest (ROIs) for DTI-based tractography of “language-essential” tracts. Tractography was performed using FSL and MRTrix.

Results: The SD task revealed highly selective activity in the TPO-L (p<0.05, FWE-corr.) when contrasted to either reverse speech or ns-SD. The contrast of SD > SC, corrected against the reverse modality, revealed strong and rather isolated activation in the triangular part of the inferior frontal gyrus as well as the frontal eye-field (p<0.05, FWE-corr.), thus qualifying the clusters as seeding ROIs for tracking the IFOF.

Conclusion: Selected contrasts of fMRI language tasks (SD > SC) are useful to identify functionally relevant language tracts and cortex regions with low recovery potential. Validation in patients is mandatory to confirm the usefulness of the suggested imaging approach for surgical planning on the single-subject level.