Great work on PANTHER—thanks for sharing it!

I would like to know how we can ensure that each WSI contains the same number of prototypes. Or is it that, for WSIs belonging to the same category, the number of prototypes can vary, and this has no impact on the classification results? For example, in LUAD, one WSI may contain only four tissue types (e.g., alveolar areas, Stroma, Necrosis.....), while another may contain more. Does such a situation exist, and would it affect the results?

Thanks in advance!