Spatially constrained ICA from a language task enables sensorimotor network extraction in patients with...

Although resting-state fMRI is a promising alternative to task-fMRI in presurgical mapping, protocols to simultaneously assess language lateralization and sensorimotor networks within a single acquisition remain limited. To evaluate the feasibility of using connectivity analysis derived from a language fMRI task to assess language lateralization and extract the sensorimotor network in presurgical patients with space-occupying lesions. In a retrospective study, 40 presurgical patients underwent a verb generation task (VGT) and a hand motor task (HMT). Single-subject spatially-constrained ICA (scICA) was performed on VGT scans to extract sensorimotor components. Sensitivity, specificity and Dice coefficient between scICA-derived scans and HMT activation maps were calculated. The effects of different variables were analyzed using ANOVA. Forty patients (mean age, 40.50 ± 13.99; 21 men) were included. Sensorimotor components could be extracted in all patients within the predefined constrained scICA framework. Using HMT activation as comparative reference, mean ipsilesional voxelwise sensitivity, specificity and Dice coefficient of scICA-derived maps were 78%, 75% and 36%, respectively. No significant differences were found between ipsilesional and contralesional hemispheres in sensitivity or specificity. Exploratory analyses suggested higher sensitivity values in the right contralesional hemisphere, higher magnetic field strength, and typical language lateralization, although most effects did not survive FDR correction. Our study supports the feasibility of extracting sensorimotor-related networks from a language task while simultaneously determining language lateralization in patients with space-occupying lesions. This approach may provide complementary presurgical information, potentially reducing acquisition time. Further validation using direct cortical stimulation and larger prospective cohorts is warranted.