The correct scale normalization for LoG should be -σ^2 instead of -σ in the multiLoG function called by blob_LoG.
In addition, the docstring of BlobLoG says:
radius is equal to σ√2
but
radius is equal to σ√N for a N-dimensional image
is more accurate.
This error is easy to fix and I can handle it on my own (please note that this will be my first PR so any guidance is much appreciated!).