Presurgical brain mapping of the language network in patients with brain tumors using resting-state fMRI: Comparison with task fMRI.

摘要: Purpose To compare language networks derived from resting-state fMRI (rs-fMRI) with task-fMRI in patients brain tumors and investigate variables that affect rs-fMRI vs concordance. Materials Methods Independent component analysis (ICA) of was performed 20, 30, 40, 50 target components (ICA20 to ICA50) identified for presenting presurgical mapping between 1/1/2009 7/1/2015. 49 were analyzed fulfilling criteria presence tumors, no prior surgery, adequate performance. Rs-vs-task-fMRI concordance measured using Dice coefficients across varying thresholds before after noise removal. Multi-thresholded coefficient volume under the surface (DiceVUS) maximum (MaxDice) calculated. One-way Analysis Variance (ANOVA) determine significance DiceVUS MaxDice four ICA order groups. Age, Sex, Handedness, Tumor Side, Size, WHO Grade, number scrubbed volumes, image intensity root mean square (iRMS), framewise displacement (FD) used as predictors VUS a linear regression. Results Artificial elevation is seen at low due noise. Noise-removed group-mean improved increased, however ANOVA demonstrated statistically significant difference Linear regression an association iRMS ICA30-50, ICA50. Conclusion Overall there moderate group level rs-vs-task network concordance, substantial subject-level variability exists; may be reliability networks. Hum Brain Mapp, 2015. © 2015 Wiley Periodicals, Inc.