censored_ratio

anri.index.censored_ratio(d, mu, censor)

Compute the MLEM ratio d / mu, with empty bins taken as censored.

A bin with no data had every pixel below the segmentation cut. Where the model predicts less than censor there, the data cannot rule it out: the bin counts as agreeing (ratio 1). Where it predicts more, it counts as censor / 2. censor = 0: empty bins are zeros (ratio 0), plain MLEM.

Return type:

Array