Negative Values in fmriprep Output

I’m working with some BOLD data that I have preprocessed with fmriprep and just realized that the spatially normalized MNI space images sometimes contain negative voxels. I looked at the input images and those have only positive values so the negatives are being introduced somewhere during preprocessing.

Is this something to be concerned about? For this participant, there are very low intensity spots inside the ventricles that then turn into negatives after running through fmriprep normalization. I’ve attached a photo to show one location with negative voxels after fmriprep processing.

Thanks in advance!!