Is DARTEL-based voxel-based morphometry affected by width of smoothing kernel and group size? A study using simulated...

Area of Science:

• Neuroimaging analysis
• Statistical modeling in neuroscience

Background:

• Voxel-based morphometry (VBM) is a key neuroimaging technique for detecting regional brain differences.
• Smoothing kernel width and group size are critical parameters influencing VBM statistical power.
• DARTEL is a novel registration method aiming to improve VBM accuracy.

Purpose of the Study:

• To evaluate the impact of the DARTEL registration method on the required smoothing kernel width in VBM.
• To determine the minimum participant group size for reliable VBM findings.
• To investigate the interplay between kernel width, group size, and detection accuracy.

Main Methods:

• Simulated atrophy was used to assess VBM detection accuracy.
• Group sizes of 10, 15, 25, and 50 were compared.
• Smoothing kernel widths ranged from 0 to 12 mm.

Main Results:

• A 6 mm kernel yielded highest accuracy for groups of 50, while 8-10 mm was optimal for groups of 25 (familywise correction).
• A group size of 25 was the minimum for cross-sectional comparisons.
• A group size of 15 was sufficient for longitudinal comparisons (false discovery rate correction).

Conclusions:

• DARTEL-based VBM benefits from smaller smoothing kernels, particularly with larger group sizes.
• The optimal kernel size is influenced by group size and statistical thresholding.
• Careful consideration of kernel selection relative to group size and statistical criteria is essential for robust VBM findings.