Create good voxel mask

[wiheto]

Relates to:

[PROC] With the purpose of filtering out ROIs with an insufficient signal-to-noise ratio (SNR), the following procedure will be applied. For each subject, the average BOLD intensity of all grey matter voxels will be classified with a Gaussian mixture model (GMM) using 2-5 Gaussians (fit dependent per subject, manually evaluated). The dropout signal due to the noise will have a lower average BOLD intensity compared to the voxels containing signal. All voxels that are higher in average signal intensity than the 5th percentile of the Gaussian with the largest mean from the GMM will be marked as “good voxels”. This will create “good voxel masks”. If an ROI has less than 50% of its voxels marked as good, then that ROI will be excluded from the analysis (masked out).

• [x] create avg voxel intensity
• [x] GMMs
• [x] manually control GMM
• [x] check 50% voxel threshold per ROI
• [x] make parcellation

[wiheto]

Done in #26