FMRI-based Presurgical Mapping For Eloquent Language Areas Using Healthy Cohort

Presurgical localization of eloquent motor and language areas can dramatically decrease the risk of post-surgical complications, and improve the life quality of the patients with brain tumor. As a non-invasive imaging technique, functional magnetic resonance imaging(fMRI) is gradually being implemented in the presurgical language mapping. Block-design task paradigm for fMRI has already been a conventional technique, while resting-state fMRI-based language mapping is a novel one.This study adopted a newly developed fMRI experimental paradigm, the “state”-related fMRI(we specifically used slow event-related paradigm, called “new paradigm” from here on), to locate language areas(especially the language production-related areas), and tried to compare this new paradigm and its corresponding analysis method with the conventional ones. We hypothesized that the new paradigm with the corresponding analysis might provide more accurate guidance for pre-operative planning.We collected fMRI data of 51 healthy subjects, who participated in a blocked fMRI(i.e., conventional language localization task), a slow event-related fMRI(“new paradigm”) and a resting-state fMRI scan. With the comparisons of regional functional connectivity(using regional homogeneity [ReHo] as an index) results as well as those of language functional network results(using seed based functional connectivity [FC] and independent component analysis [ICA]) obtained from the new paradigm and those from the resting state, we found an increased regional FC under the new paradigm compared with resting state, which indicated better participation in language functions during the new paradigm than resting state. We also found that compared with resting-state result, the sensitivity and specificity of language network localization using FC were better for the new paradigm; while such a difference was not significant if using ICA. By checking the back-reconstructed ICA results of every subject, we proved that language areas could be located either by the combination of resting-state fMRI and ICA or by the combination of slow event-related fMRI and ICA. Language localization results from the new paradigm also outperformed the conventional blocked fMRI-based ones. Specifically, the localization of language areas had better specificity under the resting state and the new paradigm, and the results from both paradigms were comparable, further proved by each other; the task activation results was suboptimal, as some irrelevant areas were also co-activated, which were not putative language areas.We demonstrated that the language localization results from these two “state”-based fMRI paradigms were comparable. When combined with FC or ICA, the slow event-related f MRI could be a supplement to block fMRI-based results. This study was the first one which demonstrated the feasibility of the “state”-based fMRI paradigm in pre-operative language mapping. The slow event-related language task paradigm has a better controllability and a lower demand on patients. ICA results from the new paradigm were comparable with those from resting-state, which can be applied to pre-operative localization for tumor patients. This paper also proved for the first time that the language network extracted by ICA during resting-state has similar spatial pattern to that activated during language task, which demonstrated the feasibility of language localization by using pure resting-state fMRI for the subjects who cannot perform language-related tasks.