Hello,
I have two questions regarding inputting an image file with a single ROI, thresholded by intensity (e.g., from ICA) but NOT binarized:
1. Will the resulting time-series for this ROI be weighted appropriately such that the most intense voxels contribute more to the ROI's time course than the ROI's least intense voxels? Or does CONN force binarizing on the ROI voxels upon import (weighting all voxels above some threshold the same)?
2. I know that people typically use binarized ROIs, but is a non-binarized ROI approach all that undesirable? I would like to construct separate ROIs from a single independent component via thresholding, but I would like to conserve the weighting for each ROI rather than binarize all ROIs. I understand that in practice, this might have to be done outside of the CONN framework, but it would be helpful to get y'alls take on the matter before I go too far down this path.
Best,
Scott
I have two questions regarding inputting an image file with a single ROI, thresholded by intensity (e.g., from ICA) but NOT binarized:
1. Will the resulting time-series for this ROI be weighted appropriately such that the most intense voxels contribute more to the ROI's time course than the ROI's least intense voxels? Or does CONN force binarizing on the ROI voxels upon import (weighting all voxels above some threshold the same)?
2. I know that people typically use binarized ROIs, but is a non-binarized ROI approach all that undesirable? I would like to construct separate ROIs from a single independent component via thresholding, but I would like to conserve the weighting for each ROI rather than binarize all ROIs. I understand that in practice, this might have to be done outside of the CONN framework, but it would be helpful to get y'alls take on the matter before I go too far down this path.
Best,
Scott