Open poldrack opened 2 years ago
Is it worth making a third suffix for non probabilistic/partial-volume segmentations? I don't feel very strongly here, but if the downstream processing could depend on whether you have a probabilistic atlas or an ICA-based atlas, it might be worth indicating it in the suffix.
it might be best to just create a new suffix for non-probabilistic continuous segmentations, since that won't break the existing scheme.
Based on a discussion with @Lestropie et al at the BIDS connectivity meeting:
Currently the common derivatives standard describes a distinction between "discrete" and "probabilistic" parcellations. However, there will be use cases (e.g. ICA/PCA) where one may want to use a map that is continuous but not a proper probability. It might be useful to change the language from "probabilistic" to "continuous" since it doesn't seem that the probabilistic part is essential, and this would support a much broader set of use cases.