Cluster AMICO kernels computation #21
Description
Description of feature
This might lead more to an extended discussion instead of a simple feature request, but I'm listing it here so we can come back to it later.
Previous flows (i.e. noddi_flow & freewater_flow) assumed that the input data had the same acquisition protocols with the same b-values. They were computing the AMICO/commit kernels based on only the first subject and they were using them with all other subjects.
To avoid this assumption on the input data, I implemented the kernels to be calculated for every subject since we don't have the tools ready yet to compare and cluster the input data based on the acquisition protocols. However, this could be an interesting optimization that we could bring in the future to avoid computing the same kernels several times.