Sparse Sampling of Silence Type I Errors With an Emphasis on Primary Auditory Cortex

Area of Science:

• Neuroimaging
• Functional Magnetic Resonance Imaging (fMRI)

Background:

• Sparse sampling functional MRI (ssfMRI) enhances the primary auditory cortex blood oxygen level-dependent (BOLD) signal by interspersing silent periods, mitigating scanner noise artifacts.
• Concerns exist regarding elevated type I error rates in resting-state fMRI due to hemodynamic response function (HRF) modeling techniques, potentially leading to unacceptable false positive findings.

Purpose of the Study:

• To investigate type I error rates in sparse sampling functional MRI (ssfMRI) across whole-brain and primary auditory cortex voxel-wise activation patterns.
• To evaluate the impact of common ssfMRI analysis techniques on false positive rates.

Main Methods:

• Participants (n=15) underwent ssfMRI scans.
• An optimized paradigm determined the auditory stimuli HRF, which was then used with silent stimuli to assess false positives.
• Voxel-wise analysis was performed on both whole-brain and primary auditory cortex data.

Main Results:

• Common ssfMRI analysis techniques yield high type I error rates.
• Similar error distributions were observed for whole-brain and primary auditory cortex analyses.
• Type I error rates at P < 0.05, P < 0.01, and P < 0.001 were substantial, particularly for the auditory cortex (e.g., 9.02 ± 1.79% at P < 0.05).

Conclusions:

• Standard ssfMRI analysis methods are associated with high false positive rates.
• A conservative alpha level (e.g., P < 0.001) is recommended for ssfMRI analyses to enhance result reliability.
• Findings highlight the need for careful statistical thresholding in ssfMRI research.