High Multiband Acceleration Degrades Resting-State Functional MRI Reliability and Signal Quality Under Anesthesia

Resting-state fMRI (rs-fMRI) is increasingly used to map brain networks in patients under anesthesia, but technical factors can affect its utility. We evaluated the effects of sevoflurane, multiband acceleration, and scan duration on rs-fMRI signal quality and within-subject reliability under anesthesia.

We retrospectively analyzed 64 clinical rs-fMRI scans acquired under anesthesia, with or without sevoflurane and multiband factor 5 acceleration. Temporal signal-to-noise ratio (tSNR) was used as a measure of signal quality. For each patient, the scan was split in half, and seed-based connectivity maps were generated for the primary motor cortex (M1), posterior cingulate cortex (PCC), and subgenual anterior cingulate cortex (sgACC). Split-half spatial correlations were used to assess within-subject reliability. Group comparisons examined differences in tSNR and reliability across conditions, and correlations with scan duration were tested.

Multiband acceleration was associated with significantly lower tSNR (U = 652.0, p = 8.9 × 10─⁶) and reduced split-half reliability for M1 (p = 0.019), PCC (p = 0.010), and sgACC (p = 0.0064). Sevoflurane showed no significant effect on tSNR or reliability. Longer scan duration correlated with improved reliability for M1 (r = 0.38, p = 0.003) but not for PCC or sgACC. No correlation was found between tSNR and reliability.

Hight multiband acceleration reduces both signal quality and reliability of rs-fMRI under anesthesia. Sevoflurane had no measurable effect. The lack of correlation between tSNR and reliability underscores the need for more robust metrics when evaluating scan quality.