search

fiji / Colocalisation_Analysis

Fiji's plugin for colocalization analysis
http://imagej.net/Coloc_2
GNU General Public License v3.0
24 stars 18 forks source link

Costes randomization occurs over bounding box of ROI instead of ROI region itself #75

Open andmccall opened 2 years ago

andmccall commented 2 years ago

Randomization occurs over the bounding box of the ROI regardless of which selection method is used (normal ROI, mask, or ROI manager). This can have a significant difference on the results, leading to more false positives. Can be tested using the plugin itself by checking the "Display Shuffled images" button.

Also, I noticed that Coloc2 applies a smoothing function to the randomized data. In my testing on my plugin I noticed that smoothing also leads to more false positives. Might be worth looking into.

imagesc-bot commented 2 years ago

This issue has been mentioned on Image.sc Forum. There might be relevant details there:

https://forum.image.sc/t/coloc-2-threshold-varies-widely-across-timepoints/69429/7

andmccall commented 2 years ago

I should probably mention too that I have a functioning costes randomizer class that only randomizes within the defined region, and that you could probably integrate with Coloc2. I'm not the best programmer, so it probably needs some polishing:

https://github.com/andmccall/Colocalization_by_Cross_Correlation/blob/master/src/main/java/CostesRandomizer.java

ctrueden commented 2 years ago

Thanks @andmccall, I appreciate it! I will try to come around to this issue when I can, but my queue is rather absurdly long. Previous developers of Coloc 2 including @tomka and @etadobson are not on the project anymore, so it will fall to me, unless someone in the community is able to file a PR to fix this issue in the meantime.