Check junction properties

[wgiese]

(1) junction countour length = perimeter (2) junction protein area = otsu threshold of junction img x dilated outline mask (3) fragmented junction contour = junction protein area x cellpose outline mask everything per cell label (4) junction protein intensity = marker_sum_expression_mem

Secondary: data frame operations: (i) coverage_index = fragemented junction countour/ perimeter (ii) interface occupancy = fragmented junction area/ dilated outline mask area (iii) intensity_per_interface_area = marker_mean_expr_mem (iv) cluster_density = junction protein intensity/junction protein area

Polarity measures: (a) polarity junction marker expression = similar to marker polarity : marker polarity applied on dilated outline mask x junction img (b) polarity junction marker expression = similar to marker polarity : marker polarity applied on dilated outline mask x junction protein area

[wgiese]

rename: mem -> junction

[jpalbrecht]

• make sure properties are correctly implemented.

[wgiese]

check again perimeter calculation, it seems that cell and nuclei perimeters are unreasonably large compared to areas. A higher accuracy might be needed, see https://scikit-image.org/docs/stable/auto_examples/segmentation/plot_perimeters.html

[jpalbrecht]

fragmented junction contour:

• adapt perimeter calculation for discontinous parts
• how does junction mapper solve this?
• not each pixel counts equally
• solving: how much does each pixel contribute to the perimeter?

-> put this feature to the long bench. This can take time to implement as well as increase runtime.